path: root/src/analyzer.js

//"use strict";

// Analyze intertype data. Calculates things that are necessary in order
// to do the final conversion into JavaScript later, for example,
// properties of variables, loop structures of functions, etc.

var VAR_NATIVE = 'native';
var VAR_NATIVIZED = 'nativized';
var VAR_EMULATED = 'emulated';

var ENTRY_IDENT = toNiceIdent('%0');

function recomputeLines(func) {
  func.lines = func.labels.map(function(label) { return label.lines }).reduce(concatenator, []);
}

// Handy sets

var BRANCH_INVOKE = set('branch', 'invoke');
var LABEL_ENDERS = set('branch', 'return', 'switch');
var SIDE_EFFECT_CAUSERS = set('call', 'invoke', 'atomic');
var UNUNFOLDABLE = set('value', 'structvalue', 'type', 'phiparam');
var SHADOW_FLIP = { i64: 'double', double: 'i64' }; //, i32: 'float', float: 'i32' };

// Analyzer

function analyzer(data, sidePass) {
  var mainPass = !sidePass;

  // Substrate
  var substrate = new Substrate('Analyzer');

  // Sorter
  substrate.addActor('Sorter', {
    processItem: function(item) {
      item.items.sort(function (a, b) { return a.lineNum - b.lineNum });
      this.forwardItem(item, 'Gatherer');
    }
  });

  // Gatherer
  substrate.addActor('Gatherer', {
    processItem: function(item) {
      // Single-liners
      ['globalVariable', 'functionStub', 'unparsedFunction', 'unparsedGlobals', 'unparsedTypes', 'alias'].forEach(function(intertype) {
        var temp = splitter(item.items, function(item) { return item.intertype == intertype });
        item.items = temp.leftIn;
        item[intertype + 's'] = temp.splitOut;
      });
      var temp = splitter(item.items, function(item) { return item.intertype == 'type' });
      item.items = temp.leftIn;
      temp.splitOut.forEach(function(type) {
        //dprint('types', 'adding defined type: ' + type.name_);
        Types.types[type.name_] = type;
        if (QUANTUM_SIZE === 1) {
          Types.fatTypes[type.name_] = copy(type);
        }
      });

      // Functions & labels
      item.functions = [];
      var currLabelFinished; // Sometimes LLVM puts a branch in the middle of a label. We need to ignore all lines after that.
      item.items.sort(function(a, b) { return a.lineNum - b.lineNum });
      for (var i = 0; i < item.items.length; i++) {
        var subItem = item.items[i];
        assert(subItem.lineNum);
        if (subItem.intertype == 'function') {
          item.functions.push(subItem);
          subItem.endLineNum = null;
          subItem.lines = []; // We will fill in the function lines after the legalizer, since it can modify them
          subItem.labels = [];
          subItem.forceEmulated = false;

          // no explicit 'entry' label in clang on LLVM 2.8 - most of the time, but not all the time! - so we add one if necessary
          if (item.items[i+1].intertype !== 'label') {
            item.items.splice(i+1, 0, {
              intertype: 'label',
              ident: ENTRY_IDENT,
              lineNum: subItem.lineNum + '.5'
            });
          }
        } else if (subItem.intertype == 'functionEnd') {
          item.functions.slice(-1)[0].endLineNum = subItem.lineNum;
        } else if (subItem.intertype == 'label') {
          item.functions.slice(-1)[0].labels.push(subItem);
          subItem.lines = [];
          currLabelFinished = false;
        } else if (item.functions.length > 0 && item.functions.slice(-1)[0].endLineNum === null) {
          // Internal line
          if (!currLabelFinished) {
            item.functions.slice(-1)[0].labels.slice(-1)[0].lines.push(subItem); // If this line fails, perhaps missing a label?
            if (subItem.intertype in LABEL_ENDERS) {
              currLabelFinished = true;
            }
          } else {
            print('// WARNING: content after a branch in a label, line: ' + subItem.lineNum);
          }
        } else {
          throw 'ERROR: what is this? ' + dump(subItem);
        }
      }
      delete item.items;
      this.forwardItem(item, 'CastAway');
    }
  });

  // CastAway - try to remove bitcasts of double<-->i64, which LLVM sometimes generates unnecessarily
  // (load a double, convert to i64, use as i64).
  // We optimize this by checking if there are such bitcasts. If so we create a shadow
  // variable that is of the other type, and use that in the relevant places. (As SSA, this is valid, and
  // variable elimination later will remove the double load if it is no longer needed.)
  //
  // Note that aside from being an optimization, this is needed for correctness in some cases: If code
  // assumes it can bitcast a double to an i64 and back and forth without loss, that may be violated
  // due to NaN canonicalization.
  substrate.addActor('CastAway', {
    processItem: function(item) {
      this.forwardItem(item, 'Legalizer');
      if (USE_TYPED_ARRAYS != 2) return;

      item.functions.forEach(function(func) {
        var has = false;
        func.labels.forEach(function(label) {
          var lines = label.lines;
          for (var i = 0; i < lines.length; i++) {
            var line = lines[i];
            if (line.intertype == 'bitcast' && line.type in SHADOW_FLIP) {
              has = true;
            }
          }
        });
        if (!has) return;
        // there are integer<->floating-point bitcasts, create shadows for everything
        var shadowed = {};
        func.labels.forEach(function(label) {
          var lines = label.lines;
          var i = 0;
          while (i < lines.length) {
          var lines = label.lines;
            var line = lines[i];
            if (line.intertype == 'load' && line.type in SHADOW_FLIP) {
              if (line.pointer.intertype != 'value') { i++; continue } // TODO
              shadowed[line.assignTo] = 1;
              var shadow = line.assignTo + '$$SHADOW';
              var flip = SHADOW_FLIP[line.type];
              lines.splice(i + 1, 0, { // if necessary this element will be legalized in the next phase
                tokens: null,
                indent: 2,
                lineNum: line.lineNum + 0.5,
                assignTo: shadow,
                intertype: 'load',
                pointerType: flip + '*',
                type: flip,
                valueType: flip,
                pointer: {
                 intertype: 'value',
                 ident: line.pointer.ident,
                 type: flip + '*'
                },
                align: line.align,
                ident: line.ident
              });
              // note: no need to update func.lines, it is generated in a later pass
              i++;
            }
            i++;
          }
        });
        // use shadows where possible
        func.labels.forEach(function(label) {
          var lines = label.lines;
          for (var i = 0; i < lines.length; i++) {
            var line = lines[i];
            if (line.intertype == 'bitcast' && line.type in SHADOW_FLIP && line.ident in shadowed) {
              var shadow = line.ident + '$$SHADOW';
              line.params[0].ident = shadow;
              line.params[0].type = line.type;
              line.type2 = line.type;
            }
          }
        });
      });
    }
  });

  // Legalize LLVM unrealistic types into realistic types.
  //
  // With full LLVM optimizations, it can generate types like i888 which do not exist in
  // any actual hardware implementation, but are useful during optimization. LLVM then
  // legalizes these types into real ones during code generation. Sadly, there is no LLVM
  // IR pass to legalize them, which would have been useful and nice from a design perspective.
  // The LLVM community is also not interested in receiving patches to implement that
  // functionality, since it would duplicate existing code from the code generation
  // component. Therefore, we implement legalization here in Emscripten.
  //
  // Currently we just legalize completely unrealistic types into bundles of i32s, and just
  // the most common instructions that can be involved with such types: load, store, shifts,
  // trunc and zext.
  substrate.addActor('Legalizer', {
    processItem: function(data) {
      // Legalization
      if (USE_TYPED_ARRAYS == 2) {
        function getLegalVars(base, bits, allowLegal) {
          bits = bits || 32; // things like pointers are all i32, but show up as 0 bits from getBits
          if (allowLegal && bits <= 32) return [{ ident: base + ('i' + bits in Runtime.INT_TYPES ? '' : '$0'), bits: bits }];
          if (isNumber(base)) return getLegalLiterals(base, bits);
          if (base[0] == '{') {
            warnOnce('seeing source of illegal data ' + base + ', likely an inline struct - assuming zeroinit');
            return getLegalLiterals('0', bits);
          }
          var ret = new Array(Math.ceil(bits/32));
          var i = 0;
          if (base == 'zeroinitializer' || base == 'undef') base = 0;
          while (bits > 0) {
            ret[i] = { ident: base ? base + '$' + i : '0', bits: Math.min(32, bits) };
            bits -= 32;
            i++;
          }
          return ret;
        }
        function getLegalLiterals(text, bits) {
          var parsed = parseArbitraryInt(text, bits);
          var ret = new Array(Math.ceil(bits/32));
          var i = 0;
          while (bits > 0) {
            ret[i] = { ident: (parsed[i]|0).toString(), bits: Math.min(32, bits) }; // resign all values
            bits -= 32;
            i++;
          }
          return ret;
        }
        function getLegalStructuralParts(value) {
          return value.params.slice(0);
        }
        function getLegalParams(params, bits) {
          return params.map(function(param) {
            var value = param.value || param;
            if (isNumber(value.ident)) {
              return getLegalLiterals(value.ident, bits);
            } else if (value.intertype == 'structvalue') {
              return getLegalStructuralParts(value).map(function(part) {
                return { ident: part.ident, bits: part.type.substr(1) };
              });
            } else {
              return getLegalVars(value.ident, bits);
            }
          });
        }
        // Uses the right factor to multiply line numbers by so that they fit in between
        // the line[i] and the line after it
        function interpLines(lines, i, toAdd) {
          var prev = i >= 0 ? lines[i].lineNum : -1;
          var next = (i < lines.length-1) ? lines[i+1].lineNum : (lines[i].lineNum + 0.5);
          var factor = (next - prev)/(4*toAdd.length+3);
          for (var k = 0; k < toAdd.length; k++) {
            toAdd[k].lineNum = prev + ((k+1)*factor);
            assert(k == 0 || toAdd[k].lineNum > toAdd[k-1].lineNum);
          }
        }
        function removeAndAdd(lines, i, toAdd) {
          var item = lines[i];
          interpLines(lines, i, toAdd);
          Array.prototype.splice.apply(lines, [i, 1].concat(toAdd));
          if (i > 0) assert(lines[i].lineNum > lines[i-1].lineNum);
          if (i + toAdd.length < lines.length) assert(lines[i + toAdd.length - 1].lineNum < lines[i + toAdd.length].lineNum);
          return toAdd.length;
        }
        function legalizeFunctionParameters(params) {
          var i = 0;
          while (i < params.length) {
            var param = params[i];
            if (param.intertype == 'value' && isIllegalType(param.type)) {
              var toAdd = getLegalVars(param.ident, getBits(param.type)).map(function(element) {
                return {
                  intertype: 'value',
                  type: 'i' + element.bits,
                  ident: element.ident,
                  byval: 0
                };
              });
              Array.prototype.splice.apply(params, [i, 1].concat(toAdd));
              i += toAdd.length;
              continue;
            }
            i++;
          }
        }
        function fixUnfolded(item) {
          // Unfolded items may need some correction to work properly in the global scope
          if (item.intertype in MATHOPS) {
            item.op = item.intertype;
            item.intertype = 'mathop';
          }
        }
        data.functions.forEach(function(func) {
          // Legalize function params
          legalizeFunctionParameters(func.params);
          // Legalize lines in labels
          var tempId = 0;
          func.labels.forEach(function(label) {
            if (dcheck('legalizer')) dprint('zz legalizing: \n' + dump(label.lines));
            var i = 0, bits;
            while (i < label.lines.length) {
              var item = label.lines[i];
              var value = item;
              // Check if we need to legalize here, and do some trivial legalization along the way
              var isIllegal = false;
              walkInterdata(item, function(item) {
                if (item.intertype == 'getelementptr' || (item.intertype == 'call' && item.ident in LLVM.INTRINSICS_32)) {
                  // Turn i64 args into i32
                  for (var i = 0; i < item.params.length; i++) {
                    if (item.params[i].type == 'i64') item.params[i].type = 'i32';
                  }
                } else if (item.intertype == 'inttoptr') {
                  var input = item.params[0];
                  if (input.type == 'i64') input.type = 'i32'; // inttoptr can only care about 32 bits anyhow since pointers are 32-bit
                }
                if (isIllegalType(item.valueType) || isIllegalType(item.type)) {
                  isIllegal = true;
                } else if ((item.intertype == 'load' || item.intertype == 'store') && isStructType(item.valueType)) {
                  isIllegal = true; // storing an entire structure is illegal
                } else if (item.intertype == 'mathop' && item.op == 'trunc' && isIllegalType(item.params[1].ident)) { // trunc stores target value in second ident
                  isIllegal = true;
                }
              });
              if (!isIllegal) {
                //if (dcheck('legalizer')) dprint('no need to legalize \n' + dump(item));
                i++;
                continue;
              }
              // Unfold this line. If we unfolded, we need to return and process the lines we just
              // generated - they may need legalization too
              var unfolded = [];
              walkAndModifyInterdata(item, function(subItem) {
                // Unfold all non-value interitems that we can, and also unfold all numbers (doing the latter
                // makes it easier later since we can then assume illegal expressions are always variables
                // accessible through ident$x, and not constants we need to parse then and there)
                if (subItem != item && (!(subItem.intertype in UNUNFOLDABLE) ||
                                       (subItem.intertype == 'value' && isNumber(subItem.ident) && isIllegalType(subItem.type)))) {
                  if (item.intertype == 'phi') {
                    assert(subItem.intertype == 'value' || subItem.intertype == 'structvalue', 'We can only unfold illegal constants in phis');
                    // we must handle this in the phi itself, if we unfold normally it will not be pushed back with the phi
                  } else {
                    var tempIdent = '$$etemp$' + (tempId++);
                    subItem.assignTo = tempIdent;
                    unfolded.unshift(subItem);
                    fixUnfolded(subItem);
                    return { intertype: 'value', ident: tempIdent, type: subItem.type };
                  }
                } else if (subItem.intertype == 'switch' && isIllegalType(subItem.type)) {
                  subItem.switchLabels.forEach(function(switchLabel) {
                    if (switchLabel.value[0] != '$') {
                      var tempIdent = '$$etemp$' + (tempId++);
                      unfolded.unshift({
                        assignTo: tempIdent,
                        intertype: 'value',
                        ident: switchLabel.value,
                        type: subItem.type
                      });
                      switchLabel.value = tempIdent;
                    }
                  });
                }
              });
              if (unfolded.length > 0) {
                interpLines(label.lines, i-1, unfolded);
                Array.prototype.splice.apply(label.lines, [i, 0].concat(unfolded));
                continue; // remain at this index, to unfold newly generated lines
              }
              // This is an illegal-containing line, and it is unfolded. Legalize it now
              dprint('legalizer', 'Legalizing ' + item.intertype + ' at line ' + item.lineNum);
              var finalizer = null;
              switch (item.intertype) {
                case 'store': {
                  var toAdd = [];
                  bits = getBits(item.valueType);
                  var elements = getLegalParams([item.value], bits)[0];
                  var j = 0;
                  elements.forEach(function(element) {
                    var tempVar = '$st$' + (tempId++) + '$' + j;
                    toAdd.push({
                      intertype: 'getelementptr',
                      assignTo: tempVar,
                      ident: item.pointer.ident,
                      type: '[0 x i32]*',
                      params: [
                        { intertype: 'value', ident: item.pointer.ident, type: '[0 x i32]*' }, // technically a bitcase is needed in llvm, but not for us
                        { intertype: 'value', ident: '0', type: 'i32' },
                        { intertype: 'value', ident: j.toString(), type: 'i32' }
                      ],
                    });
                    var actualSizeType = 'i' + element.bits; // The last one may be smaller than 32 bits
                    toAdd.push({
                      intertype: 'store',
                      valueType: actualSizeType,
                      value: { intertype: 'value', ident: element.ident, type: actualSizeType },
                      pointer: { intertype: 'value', ident: tempVar, type: actualSizeType + '*' },
                      ident: tempVar,
                      pointerType: actualSizeType + '*',
                      align: item.align,
                    });
                    j++;
                  });
                  Types.needAnalysis['[0 x i32]'] = 0;
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                // call, return: Return the first 32 bits, the rest are in temp
                case 'call': {
                  var toAdd = [value];
                  // legalize parameters
                  legalizeFunctionParameters(value.params);
                  // legalize return value, if any
                  var returnType = getReturnType(item.type);
                  if (value.assignTo && isIllegalType(returnType)) {
                    bits = getBits(returnType);
                    var elements = getLegalVars(item.assignTo, bits);
                    // legalize return value
                    value.assignTo = elements[0].ident;
                    for (var j = 1; j < elements.length; j++) {
                      var element = elements[j];
                      toAdd.push({
                        intertype: 'value',
                        assignTo: element.ident,
                        type: element.bits,
                        ident: 'tempRet' + (j - 1)
                      });
                      assert(j<10); // TODO: dynamically create more than 10 tempRet-s
                    }
                  }
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'landingpad': {
                  // not much to legalize
                  i++;
                  continue;
                }
                case 'return': {
                  bits = getBits(item.type);
                  var elements = getLegalVars(item.value.ident, bits);
                  item.value.ident = '(';
                  for (var j = 1; j < elements.length; j++) {
                    item.value.ident += 'tempRet' + (j-1) + '=' + elements[j].ident + ',';
                  }
                  item.value.ident += elements[0].ident + ')';
                  i++;
                  continue;
                }
                case 'invoke': {
                  legalizeFunctionParameters(value.params);
                  // We can't add lines after this, since invoke already modifies control flow. So we handle the return in invoke
                  i++;
                  continue;
                }
                case 'value': {
                  bits = getBits(value.type);
                  var elements = getLegalVars(item.assignTo, bits);
                  var values = getLegalLiterals(item.ident, bits);
                  var j = 0;
                  var toAdd = elements.map(function(element) {
                    return {
                      intertype: 'value',
                      assignTo: element.ident,
                      type: 'i' + bits,
                      ident: values[j++].ident
                    };
                  });
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'structvalue': {
                  bits = getBits(value.type);
                  var elements = getLegalVars(item.assignTo, bits);
                  var toAdd = [];
                  for (var j = 0; j < item.params.length; j++) {
                    toAdd[j] = {
                      intertype: 'value',
                      assignTo: elements[j].ident,
                      type: 'i32',
                      ident: item.params[j].ident
                    };
                  }
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'load': {
                  bits = getBits(value.valueType);
                  var elements = getLegalVars(item.assignTo, bits);
                  var j = 0;
                  var toAdd = [];
                  elements.forEach(function(element) {
                    var tempVar = '$ld$' + (tempId++) + '$' + j;
                    toAdd.push({
                      intertype: 'getelementptr',
                      assignTo: tempVar,
                      ident: value.pointer.ident,
                      type: '[0 x i32]*',
                      params: [
                        { intertype: 'value', ident: value.pointer.ident, type: '[0 x i32]*' }, // technically bitcast is needed in llvm, but not for us
                        { intertype: 'value', ident: '0', type: 'i32' },
                        { intertype: 'value', ident: j.toString(), type: 'i32' }
                      ]
                    });
                    var newItem = {
                      intertype: 'load',
                      assignTo: element.ident,
                      pointerType: 'i32*',
                      valueType: 'i32',
                      type: 'i32',
                      pointer: { intertype: 'value', ident: tempVar, type: 'i32*' },
                      ident: tempVar,
                      align: value.align
                    };
                    var newItem2 = null;
                    // The last one may be smaller than 32 bits
                    if (element.bits < 32) {
                      newItem.assignTo += '$preadd$';
                      newItem2 = {
                        intertype: 'mathop',
                        op: 'and',
                        assignTo: element.ident,
                        type: 'i32',
                        params: [{
                          intertype: 'value',
                          type: 'i32',
                          ident: newItem.assignTo
                        }, {
                          intertype: 'value',
                          type: 'i32',
                          ident: (0xffffffff >>> (32 - element.bits)).toString()
                        }],
                      };
                    }
                    toAdd.push(newItem);
                    if (newItem2) toAdd.push(newItem2);
                    j++;
                  });
                  Types.needAnalysis['[0 x i32]'] = 0;
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'phi': {
                  bits = getBits(value.type);
                  var toAdd = [];
                  var elements = getLegalVars(item.assignTo, bits);
                  var j = 0;
                  var values = getLegalParams(value.params, bits);
                  elements.forEach(function(element) {
                    var k = 0;
                    toAdd.push({
                      intertype: 'phi',
                      assignTo: element.ident,
                      type: 'i' + element.bits,
                      params: value.params.map(function(param) {
                        return {
                          intertype: 'phiparam',
                          label: param.label,
                          value: {
                           intertype: 'value',
                           ident: values[k++][j].ident,
                           type: 'i' + element.bits,
                          }
                        };
                      })
                    });
                    j++;
                  });
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'switch': {
                  i++;
                  continue; // special case, handled in makeComparison
                }
                case 'va_arg': {
                  assert(value.type == 'i64');
                  assert(value.value.type == 'i32*', value.value.type);
                  i += removeAndAdd(label.lines, i, range(2).map(function(x) {
                    return {
                      intertype: 'va_arg',
                      assignTo: value.assignTo + '$' + x,
                      type: 'i32',
                      value: {
                        intertype: 'value',
                        ident: value.value.ident, // We read twice from the same i32* var, incrementing // + '$' + x,
                        type: 'i32*'
                      }
                    };
                  }));
                  continue;
                }
                case 'extractvalue': { // XXX we assume 32-bit alignment in extractvalue/insertvalue,
                                       // but in theory they can run on packed structs too (see use getStructuralTypePartBits)
                  // potentially legalize the actual extracted value too if it is >32 bits, not just the extraction in general
                  var index = item.indexes[0][0].text;
                  var parts = getStructureTypeParts(item.type);
                  var indexedType = parts[index];
                  var targetBits = getBits(indexedType);
                  var sourceBits = getBits(item.type);
                  var elements = getLegalVars(item.assignTo, targetBits, true); // possibly illegal
                  var sourceElements = getLegalVars(item.ident, sourceBits); // definitely illegal
                  var toAdd = [];
                  var sourceIndex = 0;
                  for (var partIndex = 0; partIndex < parts.length; partIndex++) {
                    if (partIndex == index) {
                      for (var j = 0; j < elements.length; j++) {
                        toAdd.push({
                          intertype: 'value',
                          assignTo: elements[j].ident,
                          type: 'i' + elements[j].bits,
                          ident: sourceElements[sourceIndex+j].ident
                        });
                      }
                      break;
                    }
                    sourceIndex += getStructuralTypePartBits(parts[partIndex])/32;
                  }
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'insertvalue': {
                  var index = item.indexes[0][0].text; // the modified index
                  var parts = getStructureTypeParts(item.type);
                  var indexedType = parts[index];
                  var indexBits = getBits(indexedType);
                  var bits = getBits(item.type); // source and target
                  bits = getBits(value.type);
                  var toAdd = [];
                  var elements = getLegalVars(item.assignTo, bits);
                  var sourceElements = getLegalVars(item.ident, bits);
                  var indexElements = getLegalVars(item.value.ident, indexBits, true); // possibly legal
                  var sourceIndex = 0;
                  for (var partIndex = 0; partIndex < parts.length; partIndex++) {
                    var currNum = getStructuralTypePartBits(parts[partIndex])/32;
                    for (var j = 0; j < currNum; j++) {
                      toAdd.push({
                        intertype: 'value',
                        assignTo: elements[sourceIndex+j].ident,
                        type: 'i' + elements[sourceIndex+j].bits,
                        ident: partIndex == index ? indexElements[j].ident : sourceElements[sourceIndex+j].ident
                      });
                    }
                    sourceIndex += currNum;
                  }
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
                case 'bitcast': {
                  var inType = item.type2;
                  var outType = item.type;
                  if ((inType in Runtime.INT_TYPES && outType in Runtime.FLOAT_TYPES) ||
                      (inType in Runtime.FLOAT_TYPES && outType in Runtime.INT_TYPES)) {
                    i++;
                    continue; // special case, handled in processMathop
                  }
                  // fall through
                }
                case 'inttoptr': case 'ptrtoint': case 'zext': case 'sext': case 'trunc': case 'ashr': case 'lshr': case 'shl': case 'or': case 'and': case 'xor': {
                  value = {
                    op: item.intertype,
                    variant: item.variant,
                    type: item.type,
                    params: item.params
                  };
                  // fall through
                }
                case 'mathop': {
                  var toAdd = [];
                  var sourceBits = getBits(value.params[0].type);
                  // All mathops can be parametrized by how many shifts we do, and how big the source is
                  var shifts = 0;
                  var targetBits = sourceBits;
                  var processor = null;
                  var signed = false;
                  switch (value.op) {
                    case 'ashr': {
                      signed = true;
                      // fall through
                    }
                    case 'lshr': {
                      shifts = parseInt(value.params[1].ident);
                      break;
                    }
                    case 'shl': {
                      shifts = -parseInt(value.params[1].ident);
                      break;
                    }
                    case 'sext': {
                      signed = true;
                      // fall through
                    }
                    case 'trunc': case 'zext': case 'ptrtoint': {
                      targetBits = getBits(value.params[1] ? value.params[1].ident : value.type);
                      break;
                    }
                    case 'inttoptr': {
                      targetBits = 32;
                      break;
                    }
                    case 'bitcast': {
                      if (!sourceBits) {
                        // we can be asked to bitcast doubles or such to integers, handle that as best we can (if it's a double that
                        // was an x86_fp80, this code will likely break when called)
                        sourceBits = targetBits = Runtime.getNativeTypeSize(value.params[0].type);
                        warn('legalizing non-integer bitcast on ll #' + item.lineNum);
                      }
                      break;
                    }
                    case 'select': {
                      sourceBits = targetBits = getBits(value.params[1].type);
                      var params = getLegalParams(value.params.slice(1), sourceBits);
                      processor = function(result, j) {
                        return {
                          intertype: 'mathop',
                          op: 'select',
                          type: 'i' + params[0][j].bits,
                          params: [
                            value.params[0],
                            { intertype: 'value', ident: params[0][j].ident, type: 'i' + params[0][j].bits },
                            { intertype: 'value', ident: params[1][j].ident, type: 'i' + params[1][j].bits }
                          ]
                        };
                      };
                      break;
                    }
                    case 'or': case 'and': case 'xor': case 'icmp': {
                      var otherElements = getLegalVars(value.params[1].ident, sourceBits);
                      processor = function(result, j) {
                        return {
                          intertype: 'mathop',
                          op: value.op,
                          variant: value.variant,
                          type: 'i' + otherElements[j].bits,
                          params: [
                            result,
                            { intertype: 'value', ident: otherElements[j].ident, type: 'i' + otherElements[j].bits }
                          ]
                        };
                      };
                      if (value.op == 'icmp') {
                        if (sourceBits == 64) { // handle the i64 case in processMathOp, where we handle full i64 math
                          i++;
                          continue;
                        }
                        finalizer = function() {
                          var ident = '';
                          for (var i = 0; i < targetElements.length; i++) {
                            if (i > 0) {
                              switch(value.variant) {
                                case 'eq': ident += '&'; break;
                                case 'ne': ident += '|'; break;
                                default: throw 'unhandleable illegal icmp: ' + value.variant;
                              }
                            }
                            ident += targetElements[i].ident;
                          }
                          return {
                            intertype: 'value',
                            ident: ident,
                            type: 'rawJS',
                            assignTo: item.assignTo
                          };
                        }
                      }
                      break;
                    }
                    case 'add': case 'sub': case 'sdiv': case 'udiv': case 'mul': case 'urem': case 'srem':
                    case 'uitofp': case 'sitofp': case 'fptosi': case 'fptoui': {
                      // We cannot do these in parallel chunks of 32-bit operations. We will handle these in processMathop
                      i++;
                      continue;
                    }
                    default: throw 'Invalid mathop for legalization: ' + [value.op, item.lineNum, dump(item)];
                  }
                  // Do the legalization
                  var sourceElements = getLegalVars(value.params[0].ident, sourceBits, true);
                  if (!isNumber(shifts)) {
                    // We can't statically legalize this, do the operation at runtime TODO: optimize
                    assert(sourceBits == 64, 'TODO: handle nonconstant shifts on != 64 bits');
                    assert(PRECISE_I64_MATH, 'Must have precise i64 math for non-constant 64-bit shifts');
                    Types.preciseI64MathUsed = 1;
                    value.intertype = 'value';
                    value.ident = 'var ' + value.assignTo + '$0 = ' +
                        asmCoercion('_bitshift64' + value.op[0].toUpperCase() + value.op.substr(1) + '(' + 
                          asmCoercion(sourceElements[0].ident, 'i32') + ',' +
                          asmCoercion(sourceElements[1].ident, 'i32') + ',' +
                          asmCoercion(value.params[1].ident + '$0', 'i32') + ')', 'i32'
                        ) + ';' +
                        'var ' + value.assignTo + '$1 = tempRet0;';
                    value.assignTo = null;
                    i++;
                    continue;
                  }
                  var targetElements = getLegalVars(item.assignTo, targetBits);
                  var sign = shifts >= 0 ? 1 : -1;
                  var shiftOp = shifts >= 0 ? 'shl' : 'lshr';
                  var shiftOpReverse = shifts >= 0 ? 'lshr' : 'shl';
                  var whole = shifts >= 0 ? Math.floor(shifts/32) : Math.ceil(shifts/32);
                  var fraction = Math.abs(shifts % 32);
                  if (signed) {
                    var signedFill = '(' + makeSignOp(sourceElements[sourceElements.length-1].ident, 'i' + sourceElements[sourceElements.length-1].bits, 're', 1, 1) + ' < 0 ? -1 : 0)';
                    var signedKeepAlive = { intertype: 'value', ident: sourceElements[sourceElements.length-1].ident, type: 'i32' };
                  }
                  for (var j = 0; j < targetElements.length; j++) {
                    var result = {
                      intertype: 'value',
                      ident: (j + whole >= 0 && j + whole < sourceElements.length) ? sourceElements[j + whole].ident : (signed ? signedFill : '0'),
                      params: [(signed && j + whole > sourceElements.length) ? signedKeepAlive : null],
                      type: 'i32',
                    };
                    if (j == 0 && sourceBits < 32) {
                      // zext sign correction
                      result.ident = makeSignOp(result.ident, 'i' + sourceBits, isUnsignedOp(value.op) ? 'un' : 're', 1, 1);
                    }
                    if (fraction != 0) {
                      var other = {
                        intertype: 'value',
                        ident: (j + sign + whole >= 0 && j + sign + whole < sourceElements.length) ? sourceElements[j + sign + whole].ident : (signed ? signedFill : '0'),
                        params: [(signed && j + sign + whole > sourceElements.length) ? signedKeepAlive : null],
                        type: 'i32',
                      };
                      other = {
                        intertype: 'mathop',
                        op: shiftOp,
                        type: 'i32',
                        params: [
                          other,
                          { intertype: 'value', ident: (32 - fraction).toString(), type: 'i32' }
                        ]
                      };
                      result = {
                        intertype: 'mathop',
                        // shifting in 1s from the top is a special case
                        op: (signed && shifts >= 0 && j + sign + whole >= sourceElements.length) ? 'ashr' : shiftOpReverse,
                        type: 'i32',
                        params: [
                          result,
                          { intertype: 'value', ident: fraction.toString(), type: 'i32' }
                        ]
                      };
                      result = {
                        intertype: 'mathop',
                        op: 'or',
                        type: 'i32',
                        params: [
                          result,
                          other
                        ]
                      }
                    }
                    if (targetElements[j].bits < 32 && shifts < 0) {
                      // truncate bits that fall off the end. This is not needed in most cases, can probably be optimized out
                      result = {
                        intertype: 'mathop',
                        op: 'and',
                        type: 'i32',
                        params: [
                          result,
                          { intertype: 'value', ident: (Math.pow(2, targetElements[j].bits)-1).toString(), type: 'i32' }
                        ]
                      }
                    }
                    if (processor) {
                      result = processor(result, j);
                    }
                    result.assignTo = targetElements[j].ident;
                    toAdd.push(result);
                  }
                  if (targetBits <= 32) {
                    // We are generating a normal legal type here
                    legalValue = {
                      intertype: 'value',
                      ident: targetElements[0].ident + (targetBits < 32 ? '&' + (Math.pow(2, targetBits)-1) : ''),
                      type: 'rawJS'
                    };
                    legalValue.assignTo = item.assignTo;
                    toAdd.push(legalValue);
                  } else if (finalizer) {
                    toAdd.push(finalizer());
                  }
                  i += removeAndAdd(label.lines, i, toAdd);
                  continue;
                }
              }
              assert(0, 'Could not legalize illegal line: ' + [item.lineNum, dump(item)]);
            }
            if (dcheck('legalizer')) dprint('zz legalized: \n' + dump(label.lines));
          });
        });
      }

      // Add function lines to func.lines, after our modifications to the label lines
      data.functions.forEach(function(func) {
        func.labels.forEach(function(label) {
          func.lines = func.lines.concat(label.lines);
        });
      });
      this.forwardItem(data, 'Typevestigator');
    }
  });

  function addTypeInternal(type, data) {
    if (type.length == 1) return;
    if (Types.types[type]) return;
    if (['internal', 'hidden', 'inbounds', 'void'].indexOf(type) != -1) return;
    if (Runtime.isNumberType(type)) return;
    dprint('types', 'Adding type: ' + type);

    // 'blocks': [14 x %struct.X] etc. If this is a pointer, we need
    // to look at the underlying type - it was not defined explicitly
    // anywhere else.
    var nonPointing = removeAllPointing(type);
    var check = /^\[(\d+)\ x\ (.*)\]$/.exec(nonPointing);
    if (check && !Types.types[nonPointing]) {
      var num = parseInt(check[1]);
      num = Math.max(num, 1); // [0 x something] is used not for allocations and such of course, but
                              // for indexing - for an |array of unknown length|, basically. So we
                              // define the 'type' as having a single field. TODO: Ensure as a sanity
                              // check that we never allocate with this (either as a child structure
                              // in the analyzer, or in calcSize in alloca).
      var subType = check[2];
      addTypeInternal(subType, data); // needed for anonymous structure definitions (see below)

      // Huge structural types are represented very inefficiently, both here and in generated JS. Best to avoid them - for example static char x[10*1024*1024]; is bad, while static char *x = malloc(10*1024*1024) is fine.
      if (num >= 10*1024*1024) warnOnce('warning: very large fixed-size structural type: ' + type + ' - can you reduce it? (compilation may be slow)');
      Types.types[nonPointing] = {
        name_: nonPointing,
        fields: range(num).map(function() { return subType }),
        lineNum: '?'
      };
      // Also add a |[0 x type]| type
      var zerod = '[0 x ' + subType + ']';
      if (!Types.types[zerod]) {
        Types.types[zerod] = {
          name_: zerod,
          fields: [subType, subType], // Two, so we get the flatFactor right. We care about the flatFactor, not the size here
          lineNum: '?'
        };
      }
      return;
    }

    // anonymous structure definition, for example |{ i32, i8*, void ()*, i32 }|
    if (type[0] == '{' || type[0] == '<') {
      type = nonPointing;
      var packed = type[0] == '<';
      Types.types[type] = {
        name_: type,
        fields: splitTokenList(tokenize(type.substr(2 + packed, type.length - 4 - 2*packed)).tokens).map(function(segment) {
          return segment[0].text;
        }),
        packed: packed,
        lineNum: '?'
      };
      return;
    }

    if (isPointerType(type)) return;
    if (['['].indexOf(type) != -1) return;
    Types.types[type] = {
      name_: type,
      fields: [ 'i' + (QUANTUM_SIZE*8) ], // a single quantum size
      flatSize: 1,
      lineNum: '?'
    };
  }

  function addType(type, data) {
    addTypeInternal(type, data);
    if (QUANTUM_SIZE === 1) {
      Types.flipTypes();
      addTypeInternal(type, data);
      Types.flipTypes();
    }
  }

  // Typevestigator
  substrate.addActor('Typevestigator', {
    processItem: function(data) {
      if (sidePass) { // Do not investigate in the main pass - it is only valid to start to do so in the first side pass,
                      // which handles type definitions, and later. Doing so before the first side pass will result in
                      // making bad guesses about types which are actually defined
        for (var type in Types.needAnalysis) {
          if (type) addType(type, data);
        }
        Types.needAnalysis = {};
      }
      this.forwardItem(data, 'Typeanalyzer');
    }
  });

  // Type analyzer
  substrate.addActor('Typeanalyzer', {
    processItem: function analyzeTypes(item, fatTypes) {
      var types = Types.types;

      // 'fields' is the raw list of LLVM fields. However, we embed
      // child structures into parent structures, basically like C.
      // So { int, { int, int }, int } would be represented as
      // an Array of 4 ints. getelementptr on the parent would take
      // values 0, 1, 2, where 2 is the entire middle structure.
      // We also need to be careful with getelementptr to child
      // structures - we return a pointer to the same slab, just
      // a different offset. Likewise, need to be careful for
      // getelementptr of 2 (the last int) - it's real index is 4.
      // The benefit of this approach is inheritance -
      //    { { ancestor } , etc. } = descendant
      // In this case it is easy to bitcast ancestor to descendant
      // pointers - nothing needs to be done. If the ancestor were
      // a new slab, it would need some pointer to the outer one
      // for casting in that direction.
      // TODO: bitcasts of non-inheritance cases of embedding (not at start)
      var more = true;
      while (more) {
        more = false;
        for (var typeName in types) {
          var type = types[typeName];
          if (type.flatIndexes) continue;
          var ready = true;
          type.fields.forEach(function(field) {
            if (isStructType(field)) {
              if (!types[field]) {
                addType(field, item);
                ready = false;
              } else {
                if (!types[field].flatIndexes) {
                  ready = false;
                }
              }
            }
          });
          if (!ready) {
            more = true;
            continue;
          }

          Runtime.calculateStructAlignment(type);

          if (dcheck('types')) dprint('type (fat=' + !!fatTypes + '): ' + type.name_ + ' : ' + JSON.stringify(type.fields));
          if (dcheck('types')) dprint('                        has final size of ' + type.flatSize + ', flatting: ' + type.needsFlattening + ' ? ' + (type.flatFactor ? type.flatFactor : JSON.stringify(type.flatIndexes)));
        }
      }

      if (QUANTUM_SIZE === 1 && !fatTypes) {
        Types.flipTypes();
        // Fake a quantum size of 4 for fat types. TODO: Might want non-4 for some reason?
        var trueQuantumSize = QUANTUM_SIZE;
        Runtime.QUANTUM_SIZE = 4;
        analyzeTypes(item, true);
        Runtime.QUANTUM_SIZE = trueQuantumSize;
        Types.flipTypes();
      }

      if (!fatTypes) {
        this.forwardItem(item, 'VariableAnalyzer');
      }
    }
  });
  
  // Variable analyzer
  substrate.addActor('VariableAnalyzer', {
    processItem: function(item) {
      // Globals

      var old = item.globalVariables;
      item.globalVariables = {};
      old.forEach(function(variable) {
        variable.impl = 'emulated'; // All global variables are emulated, for now. Consider optimizing later if useful
        item.globalVariables[variable.ident] = variable;
      });

      // Function locals

      item.functions.forEach(function(func) {
        func.variables = {};

        // LLVM is SSA, so we always have a single assignment/write. We care about
        // the reads/other uses.

        // Function parameters
        func.params.forEach(function(param) {
          if (param.intertype !== 'varargs') {
            if (func.variables[param.ident]) warn('cannot have duplicate variable names: ' + param.ident); // toNiceIdent collisions?
            func.variables[param.ident] = {
              ident: param.ident,
              type: param.type,
              origin: 'funcparam',
              lineNum: func.lineNum,
              rawLinesIndex: -1
            };
          }
        });

        // Normal variables
        func.lines.forEach(function(item, i) {
          if (item.assignTo) {
            if (func.variables[item.assignTo]) warn('cannot have duplicate variable names: ' + item.assignTo); // toNiceIdent collisions?
            var variable = func.variables[item.assignTo] = {
              ident: item.assignTo,
              type: item.type,
              origin: item.intertype,
              lineNum: item.lineNum,
              rawLinesIndex: i
            };
            if (variable.origin === 'alloca') {
              variable.allocatedNum = item.allocatedNum;
            }
            if (variable.origin === 'call') {
              variable.type = getReturnType(variable.type);
            }
          }
        });

        if (QUANTUM_SIZE === 1) {
          // Second pass over variables - notice when types are crossed by bitcast

          func.lines.forEach(function(item) {
            if (item.assignTo && item.intertype === 'bitcast') {
              // bitcasts are unique in that they convert one pointer to another. We
              // sometimes need to know the original type of a pointer, so we save that.
              //
              // originalType is the type this variable is created from
              // derivedTypes are the types that this variable is cast into
              func.variables[item.assignTo].originalType = item.type2;

              if (!isNumber(item.assignTo)) {
                if (!func.variables[item.assignTo].derivedTypes) {
                  func.variables[item.assignTo].derivedTypes = [];
                }
                func.variables[item.assignTo].derivedTypes.push(item.type);
              }
            }
          });
        }

        // Analyze variable uses

        function analyzeVariableUses() {
          dprint('vars', 'Analyzing variables for ' + func.ident + '\n');

          for (vname in func.variables) {
            var variable = func.variables[vname];

            // Whether the value itself is used. For an int, always yes. For a pointer,
            // we might never use the pointer's value - we might always just store to it /
            // read from it. If so, then we can optimize away the pointer.
            variable.hasValueTaken = false;

            variable.pointingLevels = pointingLevels(variable.type);

            variable.uses = 0;
          }

          // TODO: improve the analysis precision. bitcast, for example, means we take the value, but perhaps we only use it to load/store
          var inNoop = 0;
          func.lines.forEach(function(line) {
            walkInterdata(line, function(item) {
              if (item.intertype == 'noop') inNoop++;
              if (!inNoop) {
                if (item.ident in func.variables) {
                  func.variables[item.ident].uses++;

                  if (item.intertype != 'load' && item.intertype != 'store') {
                    func.variables[item.ident].hasValueTaken = true;
                  }
                }
              }
            }, function(item) {
              if (item.intertype == 'noop') inNoop--;
            });
          });

          //if (dcheck('vars')) dprint('analyzed variables: ' + dump(func.variables));
        }

        analyzeVariableUses();

        // Decision time

        for (vname in func.variables) {
          var variable = func.variables[vname];
          var pointedType = pointingLevels(variable.type) > 0 ? removePointing(variable.type) : null;
          if (variable.origin == 'getelementptr') {
            // Use our implementation that emulates pointers etc.
            // TODO Can we perhaps nativize some of these? However to do so, we need to discover their
            //      true types; we have '?' for them now, as they cannot be discovered in the intertyper.
            variable.impl = VAR_EMULATED;
          } else if (variable.origin == 'funcparam') {
            variable.impl = VAR_EMULATED;
          } else if (variable.type == 'i64*' && USE_TYPED_ARRAYS == 2) {
            variable.impl = VAR_EMULATED;
          } else if (MICRO_OPTS && variable.pointingLevels === 0) {
            // A simple int value, can be implemented as a native variable
            variable.impl = VAR_NATIVE;
          } else if (MICRO_OPTS && variable.origin === 'alloca' && !variable.hasValueTaken &&
                     variable.allocatedNum === 1 &&
                     (Runtime.isNumberType(pointedType) || Runtime.isPointerType(pointedType))) {
            // A pointer to a value which is only accessible through this pointer. Basically
            // a local value on the stack, which nothing fancy is done on. So we can
            // optimize away the pointing altogether, and just have a native variable
            variable.impl = VAR_NATIVIZED;
          } else {
            variable.impl = VAR_EMULATED;
          }
          if (dcheck('vars')) dprint('// var ' + vname + ': ' + JSON.stringify(variable));
        }
      });
      this.forwardItem(item, 'Signalyzer');
    }
  });

  // Sign analyzer
  //
  // Analyze our variables and detect their signs. In USE_TYPED_ARRAYS == 2,
  // we can read signed or unsigned values and prevent the need for signing
  // corrections. If on the other hand we are doing corrections anyhow, then
  // we can skip this pass.
  //
  // For each variable that is the result of a Load, we look a little forward
  // to see where it is used. We only care about mathops, since only they
  // need signs.
  //
  substrate.addActor('Signalyzer', {
    processItem: function(item) {
      this.forwardItem(item, 'QuantumFixer');
      if (USE_TYPED_ARRAYS != 2 || CORRECT_SIGNS == 1) return;

      function seekIdent(item, obj) {
        if (item.ident === obj.ident) {
          obj.found++;
        }
      }

      function seekMathop(item, obj) {
        if (item.intertype === 'mathop' && obj.found && !obj.decided) {
          if (isUnsignedOp(item.op, item.variant)) {
            obj.unsigned++;
          } else {
            obj.signed++;
          }
        }
      }

      item.functions.forEach(function(func) {
        func.lines.forEach(function(line, i) {
          if (line.intertype === 'load') {
            // Floats have no concept of signedness. Mark them as 'signed', which is the default, for which we do nothing
            if (line.type in Runtime.FLOAT_TYPES) {
              line.unsigned = false;
              return;
            }
            // Booleans are always unsigned
            var data = func.variables[line.assignTo];
            if (data.type === 'i1') {
              line.unsigned = true;
              return;
            }

            var total = data.uses;
            if (total === 0) return;
            var obj = { ident: line.assignTo, found: 0, unsigned: 0, signed: 0, total: total };
            // in loops with phis, we can also be used *before* we are defined
            var j = i-1, k = i+1;
            while(1) {
              assert(j >= 0 || k < func.lines.length, 'Signalyzer ran out of space to look for sign indications for line ' + line.lineNum);
              if (j >= 0 && walkInterdata(func.lines[j], seekIdent, seekMathop, obj)) break;
              if (k < func.lines.length && walkInterdata(func.lines[k], seekIdent, seekMathop, obj)) break;
              if (obj.total && obj.found >= obj.total) break; // see comment below
              j -= 1;
              k += 1;
            }

            // unsigned+signed might be < total, since the same ident can appear multiple times in the same mathop.
            // found can actually be > total, since we currently have the same ident in a GEP (see cubescript test)
            // in the GEP item, and a child item (we have the ident copied onto the GEP item as a convenience).
            // probably not a bug-causer, but FIXME. see also a reference to this above
            // we also leave the loop above potentially early due to this. otherwise, though, we end up scanning the
            // entire function in some cases which is very slow
            assert(obj.found >= obj.total, 'Could not Signalyze line ' + line.lineNum);
            line.unsigned = obj.unsigned > 0;
            dprint('vars', 'Signalyzer: ' + line.assignTo + ' has unsigned == ' + line.unsigned + ' (line ' + line.lineNum + ')');
          }
        });
      });
    }
  });

  // Quantum fixer
  //
  // See settings.js for the meaning of QUANTUM_SIZE. The issue we fix here is,
  // to correct the .ll assembly code so that things work with QUANTUM_SIZE=1.
  //
  substrate.addActor('QuantumFixer', {
    processItem: function(item) {
      this.forwardItem(item, 'LabelAnalyzer');
      if (QUANTUM_SIZE !== 1) return;

      // ptrs: the indexes of parameters that are pointers, whose originalType is what we want
      // bytes: the index of the 'bytes' parameter
      // TODO: malloc, realloc?
      var FIXABLE_CALLS = {
        'memcpy': { ptrs: [0,1], bytes: 2 },
        'memmove': { ptrs: [0,1], bytes: 2 },
        'memset': { ptrs: [0], bytes: 2 },
        'qsort': { ptrs: [0], bytes: 2 }
      };

      function getSize(types, type, fat) {
        if (types[type]) return types[type].flatSize;
        if (fat) {
          Runtime.QUANTUM_SIZE = 4;
        }
        var ret = Runtime.getNativeTypeSize(type);
        if (fat) {
          Runtime.QUANTUM_SIZE = 1;
        }
        return ret;
      }

      function getFlatIndexes(types, type) {
        if (types[type]) return types[type].flatIndexes;
        return [0];
      }

      item.functions.forEach(function(func) {
        function getOriginalType(param) {
          function get() {
            if (param.intertype === 'value' && !isNumber(param.ident)) {
              if (func.variables[param.ident]) {
                return func.variables[param.ident].originalType || null;
              } else {
                return item.globalVariables[param.ident].originalType;
              }
            } else if (param.intertype === 'bitcast') {
              return param.params[0].type;
            } else if (param.intertype === 'getelementptr') {
              if (param.params[0].type[0] === '[') return param.params[0].type;
            }
            return null;
          }
          var ret = get();
          if (ret && ret[0] === '[') {
            var check = /^\[(\d+)\ x\ (.*)\]\*$/.exec(ret);
            assert(check);
            ret = check[2] + '*';
          }
          return ret;
        }

        func.lines.forEach(function(line) {
          // Call
          if (line.intertype === 'call') {
            var funcIdent = LibraryManager.getRootIdent(line.ident.substr(1));
            var fixData = FIXABLE_CALLS[funcIdent];
            if (!fixData) return;
            var ptrs = fixData.ptrs.map(function(ptr) { return line.params[ptr] });
            var bytes = line.params[fixData.bytes].ident;

            // Only consider original types. This assumes memcpy always has pointers bitcast to i8*
            var originalTypes = ptrs.map(getOriginalType);
            for (var i = 0; i < originalTypes.length; i++) {
              if (!originalTypes[i]) return;
            }
            originalTypes = originalTypes.map(function(type) { return removePointing(type) });
            var sizes = originalTypes.map(function(type) { return getSize(Types.types, type) });
            var fatSizes = originalTypes.map(function(type) { return getSize(Types.fatTypes, type, true) });
            // The sizes may not be identical, if we copy a descendant class into a parent class. We use
            // the smaller size in that case. However, this may also be a bug, it is hard to tell, hence a warning
            warn(dedup(sizes).length === 1, 'All sizes should probably be identical here: ' + dump(originalTypes) + ':' + dump(sizes) + ':' +
                 line.lineNum);
            warn(dedup(fatSizes).length === 1, 'All fat sizes should probably be identical here: ' + dump(originalTypes) + ':' + dump(sizes) + ':' +
                 line.lineNum);
            var size = Math.min.apply(null, sizes);
            var fatSize = Math.min.apply(null, fatSizes);
            if (isNumber(bytes)) {
              // Figure out how much to copy.
              var fixedBytes;
              if (bytes % fatSize === 0) {
                fixedBytes = size*(bytes/fatSize);
              } else if (fatSize % bytes === 0 && size % (fatSize/bytes) === 0) {
                // Assume this is a simple array. XXX We can be wrong though! See next TODO
                fixedBytes = size/(fatSize/bytes);
              } else {
                // Just part of a structure. Align them to see how many fields. Err on copying more.
                // TODO: properly generate a complete structure, including nesteds, and calculate on that
                var flatIndexes = getFlatIndexes(Types.types, originalTypes[0]).concat(size);
                var fatFlatIndexes = getFlatIndexes(Types.fatTypes, originalTypes[0]).concat(fatSize);
                var index = 0;
                var left = bytes;
                fixedBytes = 0;
                while (left > 0) {
                  left -= fatFlatIndexes[index+1] - fatFlatIndexes[index]; // note: we copy the alignment bytes too, which is unneeded
                  fixedBytes += flatIndexes[index+1] - flatIndexes[index];
                }
              }
              line.params[fixData.bytes].ident = fixedBytes;
            } else {
              line.params[fixData.bytes].intertype = 'jsvalue';
              // We have an assertion in library::memcpy() that this is round
              line.params[fixData.bytes].ident = size + '*(' + bytes + '/' + fatSize + ')';
            }
          }
        });
      });

      // 2nd part - fix hardcoded constant offsets in global constants
      values(item.globalVariables).forEach(function(variable) {
        function recurse(item) {
          if (item.contents) {
            item.contents.forEach(recurse);
          } else if (item.intertype === 'getelementptr' && item.params[0].intertype === 'bitcast' && item.params[0].type === 'i8*') {
            var originalType = removePointing(item.params[0].params[0].type);
            var fatSize = getSize(Types.fatTypes, originalType, true);
            var slimSize = getSize(Types.types, originalType, false);
            assert(fatSize % slimSize === 0);
            item.params.slice(1).forEach(function(param) {
              if (param.intertype === 'value' && isNumber(param.ident)) {
                var corrected = parseInt(param.ident)/(fatSize/slimSize);
                assert(corrected % 1 === 0);
                param.ident = corrected.toString();
              }
            });
          } else if (item.params) {
            item.params.forEach(recurse);
          }
        }
        if (!variable.external && variable.value) recurse(variable.value);
      });
    }
  });

  function operateOnLabels(line, func) {
    function process(item, id) {
      ['label', 'labelTrue', 'labelFalse', 'toLabel', 'unwindLabel', 'defaultLabel'].forEach(function(id) {
        if (item[id]) {
          func(item, id);
        }
      });
    }
    if (line.intertype in BRANCH_INVOKE) {
      process(line);
    } else if (line.intertype == 'switch') {
      process(line);
      line.switchLabels.forEach(process);
    }
  }

  // Label analyzer
  substrate.addActor('LabelAnalyzer', {
    processItem: function(item) {
      item.functions.forEach(function(func) {
        func.labelsDict = {};
        func.labelIds = {};
        func.labelIdsInverse = {};
        func.labelIdCounter = 1;
        func.labels.forEach(function(label) {
          if (!(label.ident in func.labelIds)) {
            func.labelIds[label.ident] = func.labelIdCounter++;
            func.labelIdsInverse[func.labelIdCounter-1] = label.ident;
          }
        });
        var entryIdent = func.labels[0].ident;

        // Minify label ids to numeric ids.
        func.labels.forEach(function(label) {
          label.ident = func.labelIds[label.ident];
          label.lines.forEach(function(line) {
            operateOnLabels(line, function(item, id) {
              item[id] = func.labelIds[item[id]].toString(); // strings, because we will append as we process
            });
          });
        });

        func.labels.forEach(function(label) {
          func.labelsDict[label.ident] = label;
        });

        // Correct phis
        func.labels.forEach(function(label) {
          label.lines.forEach(function(phi) {
            if (phi.intertype == 'phi') {
              for (var i = 0; i < phi.params.length; i++) {
                phi.params[i].label = func.labelIds[phi.params[i].label];
                if (VERBOSE && !phi.params[i].label) warn('phi refers to nonexistent label on line ' + phi.lineNum);
              }
            }
          });
        });

        func.lines.forEach(function(line) {
          if (line.intertype == 'indirectbr') {
            func.forceEmulated = true;
          }
        });

        function getActualLabelId(labelId) {
          if (func.labelsDict[labelId]) return labelId;
          // If not present, it must be a surprisingly-named entry (or undefined behavior, in which case, still ok to use the entry)
          labelId = func.labelIds[entryIdent];
          assert(func.labelsDict[labelId]);
          return labelId;
        }

        // Basic longjmp support, see library.js setjmp/longjmp
        var setjmp = toNiceIdent('@setjmp');
        func.setjmpTable = null;
        for (var i = 0; i < func.labels.length; i++) {
          var label = func.labels[i];
          for (var j = 0; j < label.lines.length; j++) {
            var line = label.lines[j];
            if ((line.intertype == 'call' || line.intertype == 'invoke') && line.ident == setjmp) {
              // Add a new label
              var oldLabel = label.ident;
              var newLabel = func.labelIdCounter++;
              if (!func.setjmpTable) func.setjmpTable = [];
              func.setjmpTable.push({ oldLabel: oldLabel, newLabel: newLabel, assignTo: line.assignTo });
              func.labels.splice(i+1, 0, {
                intertype: 'label',
                ident: newLabel,
                lineNum: label.lineNum + 0.5,
                lines: label.lines.slice(j+1)
              });
              func.labelsDict[newLabel] = func.labels[i+1];
              label.lines = label.lines.slice(0, j+1);
              label.lines.push({
                intertype: 'branch',
                label: toNiceIdent(newLabel),
                lineNum: line.lineNum + 0.01, // XXX legalizing might confuse this
              });
              // Correct phis
              func.labels.forEach(function(label) {
                label.lines.forEach(function(phi) {
                  if (phi.intertype == 'phi') {
                    for (var i = 0; i < phi.params.length; i++) {
                      var sourceLabelId = getActualLabelId(phi.params[i].label);
                      if (sourceLabelId == oldLabel) {
                        phi.params[i].label = newLabel;
                      }
                    }
                  }
                });
              });
            }
          }
        }
        if (func.setjmpTable) {
          func.forceEmulated = true;
          recomputeLines(func);
        }

        // Properly implement phis, by pushing them back into the branch
        // that leads to here. We will only have the |var| definition in this location.

        // First, push phis back
        func.labels.forEach(function(label) {
          label.lines.forEach(function(phi) {
            if (phi.intertype == 'phi') {
              for (var i = 0; i < phi.params.length; i++) {
                var param = phi.params[i];
                if (VERBOSE && !param.label) warn('phi refers to nonexistent label on line ' + phi.lineNum);
                var sourceLabelId = getActualLabelId(param.label);
                if (sourceLabelId) {
                  var sourceLabel = func.labelsDict[sourceLabelId];
                  var lastLine = sourceLabel.lines.slice(-1)[0];
                  assert(lastLine.intertype in LLVM.PHI_REACHERS, 'Only some can lead to labels with phis:' + [func.ident, label.ident, lastLine.intertype]);
                  if (!lastLine.phi) {
                    lastLine.phi = true;
                    assert(!lastLine.dependent);
                    lastLine.dependent = {
                      intertype: 'phiassigns',
                      params: []
                    };
                  };
                  lastLine.dependent.params.push({
                    intertype: 'phiassign',
                    ident: phi.assignTo,
                    value: param.value,
                    targetLabel: label.ident
                  });
                }
              }
              // The assign to phi is now just a var
              phi.intertype = 'var';
              phi.ident = phi.assignTo;
              phi.assignTo = null;
            }
          });
        });

        if (func.ident in NECESSARY_BLOCKADDRS) {
          Functions.blockAddresses[func.ident] = {};
          for (var needed in NECESSARY_BLOCKADDRS[func.ident]) {
            assert(needed in func.labelIds);
            Functions.blockAddresses[func.ident][needed] = func.labelIds[needed];
          }
        }
      });
      this.forwardItem(item, 'StackAnalyzer');
    }
  });

  // Stack analyzer - calculate the base stack usage
  substrate.addActor('StackAnalyzer', {
    processItem: function(data) {
      data.functions.forEach(function(func) {
        var lines = func.labels[0].lines;
        for (var i = 0; i < lines.length; i++) {
          var item = lines[i];
          if (!item.assignTo || item.intertype != 'alloca' || !isNumber(item.allocatedNum)) break;
          item.allocatedSize = func.variables[item.assignTo].impl === VAR_EMULATED ?
            calcAllocatedSize(item.allocatedType)*item.allocatedNum: 0;
          if (USE_TYPED_ARRAYS === 2) {
            // We need to keep the stack aligned
            item.allocatedSize = Runtime.forceAlign(item.allocatedSize, Runtime.STACK_ALIGN);
          }
        }
        var index = 0;
        for (var i = 0; i < lines.length; i++) {
          var item = lines[i];
          if (!item.assignTo || item.intertype != 'alloca' || !isNumber(item.allocatedNum)) break;
          item.allocatedIndex = index;
          index += item.allocatedSize;
          delete item.allocatedSize;
        }
        func.initialStack = index;
        func.otherStackAllocations = false;
        while (func.initialStack == 0) { // one-time loop with possible abort in the middle
          // If there is no obvious need for stack management, perhaps we don't need it
          // (we try to optimize that way with SKIP_STACK_IN_SMALL). However,
          // we need to note if stack allocations other than initial allocs can happen here
          // If so, we need to rewind the stack when we leave.

          // By-value params are causes of additional allocas (although we could in theory make them normal allocas too)
          func.params.forEach(function(param) {
            if (param.byVal) {
              func.otherStackAllocations = true;
            }
          });
          if (func.otherStackAllocations) break;

          // Allocas
          var finishedInitial = false;

          lines = func.lines; // We need to consider all the function lines now, not just the first label

          for (var i = 0; i < lines.length; i++) {
            var item = lines[i];
            if (!finishedInitial && (!item.assignTo || item.intertype != 'alloca' || !isNumber(item.allocatedNum))) {
              finishedInitial = true;
            }
            if (item.intertype == 'alloca' && finishedInitial) {
              func.otherStackAllocations = true;
              break;
            }
          }
          if (func.otherStackAllocations) break;

          // Varargs
          for (var i = 0; i < lines.length; i++) {
            var item = lines[i];
            if (item.intertype == 'call' && isVarArgsFunctionType(item.type)) {
              func.otherStackAllocations = true;
              break;
            }
          }
          if (func.otherStackAllocations) break;

          break;
        }
      });
      this.forwardItem(data, 'Relooper');
    }
  });

  // ReLooper - reconstruct nice loops, as much as possible
  // This is now done in the jsify stage, using compiled relooper2
  substrate.addActor('Relooper', {
    processItem: function(item) {
      function finish() {
        item.__finalResult__ = true;
        return [item];
      }
      function makeBlock(labels, entries, labelsDict, forceEmulated) {
        if (labels.length == 0) return null;
        dprint('relooping', 'prelooping: ' + entries + ',' + labels.length + ' labels');
        assert(entries && entries[0]); // need at least 1 entry

        var emulated = {
          type: 'emulated',
          id: 'B',
          labels: labels,
          entries: entries.slice(0)
        };
        return emulated;
      }
      item.functions.forEach(function(func) {
        dprint('relooping', "// relooping function: " + func.ident);
        func.block = makeBlock(func.labels, [func.labels[0].ident], func.labelsDict, func.forceEmulated);
      });

      return finish();
    }
  });

  // Data
  substrate.addItem({
    items: data
  }, 'Sorter');

  // Solve it
  return substrate.solve();
}