Convert extended string handling functions into JS library code

Author: sbc100

ChangeLog.md

@@ -34,10 +34,28 @@ See docs/process.md for more on how version tagging works.
    - addFunction
    - removeFunction
    - allocate
-  This means they won't get included in the output unless explictly required.
-  Exporting them via `EXPORTED_RUNTIME_METHODS` will continue to work.  For
-  internal usage (without exporting them) they can be added to
-  `DEFAULT_LIBRARY_FUNCS_TO_INCLUDE`. (#17370)
+   - AsciiToString
+   - stringToAscii
+   - UTF16ToString
+   - stringToUTF16
+   - lengthBytesUTF16
+   - UTF32ToString
+   - stringToUTF32
+   - lengthBytesUTF32
+   - allocateUTF8
+   - allocateUTF8OnStack
+   - writeStringToMemory
+   - writeArrayToMemory
+   - writeAsciiToMemory
+   - intArrayFromString
+   - intArrayToString
+  However they all still available by default due to a new setting called
+  `LEGACY_RUNTIME` which is enabled by default.  When `LEGACY_RUNTIME` is
+  disabled (which may be the default soon) these symbol would only be included
+  if there were explictly exported via `EXPORTED_RUNTIME_METHODS`
+  or added to `DEFAULT_LIBRARY_FUNCS_TO_INCLUDE`.
+  `LEGACY_RUNTIME` is disabled bv default in `STRICT` mode so this change
+  only effects users of `STRICT` mode.  (#17370)
 - The `run` runtime function is no longer exported by default.  It can be added
   to `EXPORTED_RUNTIME_METHODS` if needed.
 - The getWasmTableEntry/setWasmTableEntry library function are no longer

emcc.py

@@ -1793,6 +1793,7 @@ def phase_linker_setup(options, state, newargs, user_settings):
 
   if settings.STRICT:
     default_setting(user_settings, 'STRICT_JS', 1)
+    default_setting(user_settings, 'LEGACY_RUNTIME', 1)
     default_setting(user_settings, 'AUTO_JS_LIBRARIES', 0)
     default_setting(user_settings, 'AUTO_NATIVE_LIBRARIES', 0)
     default_setting(user_settings, 'AUTO_ARCHIVE_INDEXES', 0)
@@ -2132,7 +2133,9 @@ def phase_linker_setup(options, state, newargs, user_settings):
         '$jsStackTrace',
         '$stackTrace',
         # Called by `callMain` to handle exceptions
-        '$handleException'
+        '$handleException',
+        # Needed by ccall (remove this once ccall itself is a library function)
+        '$writeArrayToMemory',
       ]
 
     if not settings.STANDALONE_WASM and (settings.EXIT_RUNTIME or settings.ASSERTIONS):
@@ -2873,6 +2876,8 @@ def phase_emscript(options, in_wasm, wasm_target, memfile):
 
   if embed_memfile():
     settings.SUPPORT_BASE64_EMBEDDING = 1
+    # _read in shell.js depends on intArrayToString when SUPPORT_BASE64_EMBEDDING is set
+    settings.DEFAULT_LIBRARY_FUNCS_TO_INCLUDE.append('$intArrayToString')
 
   emscripten.run(in_wasm, wasm_target, final_js, memfile)
   save_intermediate('original')

emscripten.py

@@ -152,6 +152,9 @@ def update_settings_glue(wasm_file, metadata):
   # When using dynamic linking the main function might be in a side module.
   # To be safe assume they do take input parametes.
   settings.MAIN_READS_PARAMS = metadata['mainReadsParams'] or bool(settings.MAIN_MODULE)
+  if settings.MAIN_READS_PARAMS and not settings.STANDALONE_WASM:
+    # callMain depends on this library function
+    settings.DEFAULT_LIBRARY_FUNCS_TO_INCLUDE += ['$allocateUTF8OnStack']
 
   if settings.STACK_OVERFLOW_CHECK and not settings.SIDE_MODULE:
     settings.EXPORTED_RUNTIME_METHODS += ['writeStackCookie', 'checkStackCookie']

src/embind/embind.js

@@ -784,10 +784,8 @@ var LibraryEmbind = {
   _embind_register_std_wstring__sig: 'vppp',
   _embind_register_std_wstring__deps: [
     '$readLatin1String', '$registerType', '$simpleReadValueFromPointer',
-#if MINIMAL_RUNTIME
     '$UTF16ToString', '$stringToUTF16', '$lengthBytesUTF16',
     '$UTF32ToString', '$stringToUTF32', '$lengthBytesUTF32',
-#endif
     ],
   _embind_register_std_wstring: function(rawType, charSize, name) {
     name = readLatin1String(name);

src/jsifier.js

@@ -506,8 +506,6 @@ function ${name}(${args}) {
 
     if (!MINIMAL_RUNTIME) {
       print('var ASSERTIONS = ' + !!ASSERTIONS + ';\n');
-
-      print(preprocess(read('arrayUtils.js')));
     }
 
     if ((SUPPORT_BASE64_EMBEDDING || FORCE_FILESYSTEM) && !MINIMAL_RUNTIME) {

src/library.js

@@ -565,11 +565,7 @@ mergeInto(LibraryManager.library, {
   tzset_impl__internal: true,
   tzset_impl__proxy: 'sync',
   tzset_impl__sig: 'viii',
-  tzset_impl__deps: [
-#if MINIMAL_RUNTIME
-    '$allocateUTF8'
-#endif
-  ],
+  tzset_impl__deps: ['$allocateUTF8'],
   tzset_impl: function(timezone, daylight, tzname) {
     var currentYear = new Date().getFullYear();
     var winter = new Date(currentYear, 0, 1);
@@ -654,10 +650,8 @@ mergeInto(LibraryManager.library, {
 
   // Note: this is not used in STANDALONE_WASM mode, because it is more
   //       compact to do it in JS.
-  strftime__deps: ['_isLeapYear', '_arraySum', '_addDays', '_MONTH_DAYS_REGULAR', '_MONTH_DAYS_LEAP'
-#if MINIMAL_RUNTIME
-    , '$intArrayFromString', '$writeArrayToMemory'
-#endif
+  strftime__deps: ['_isLeapYear', '_arraySum', '_addDays', '_MONTH_DAYS_REGULAR', '_MONTH_DAYS_LEAP',
+                   '$intArrayFromString', '$writeArrayToMemory'
   ],
   strftime__sig: 'ppppp',
   strftime: function(s, maxsize, format, tm) {
@@ -953,11 +947,8 @@ mergeInto(LibraryManager.library, {
     return _strftime(s, maxsize, format, tm); // no locale support yet
   },
 
-  strptime__deps: ['_isLeapYear', '_arraySum', '_addDays', '_MONTH_DAYS_REGULAR', '_MONTH_DAYS_LEAP', '$jstoi_q'
-#if MINIMAL_RUNTIME
-    , '$intArrayFromString'
-#endif
-  ],
+  strptime__deps: ['_isLeapYear', '_arraySum', '_addDays', '_MONTH_DAYS_REGULAR', '_MONTH_DAYS_LEAP',
+                   '$jstoi_q', '$intArrayFromString' ],
   strptime__sig: 'pppp',
   strptime: function(buf, format, tm) {
     // char *strptime(const char *restrict buf, const char *restrict format, struct tm *restrict tm);
@@ -2078,11 +2069,7 @@ mergeInto(LibraryManager.library, {
     list: [],
     map: {}
   },
-  setprotoent__deps: ['$Protocols'
-#if MINIMAL_RUNTIME
-    , '$writeAsciiToMemory'
-#endif
-  ],
+  setprotoent__deps: ['$Protocols', '$writeAsciiToMemory'],
   setprotoent: function(stayopen) {
     // void setprotoent(int stayopen);
 
@@ -2771,13 +2758,7 @@ mergeInto(LibraryManager.library, {
 
   // Look up the function name from our stack frame cache with our PC representation.
 #if USE_OFFSET_CONVERTER
-  emscripten_pc_get_function__deps: [
-    '$UNWIND_CACHE',
-    'free',
-#if MINIMAL_RUNTIME
-    '$allocateUTF8',
-#endif
-  ],
+  emscripten_pc_get_function__deps: ['$UNWIND_CACHE', 'free', '$allocateUTF8'],
   // Don't treat allocation of _emscripten_pc_get_function.ret as a leak
   emscripten_pc_get_function__noleakcheck: true,
   emscripten_pc_get_function__sig: 'pp',
@@ -2838,11 +2819,7 @@ mergeInto(LibraryManager.library, {
   },
 
   // Look up the file name from our stack frame cache with our PC representation.
-  emscripten_pc_get_file__deps: ['$convertPCtoSourceLocation', 'free',
-#if MINIMAL_RUNTIME
-    '$allocateUTF8',
-#endif
-  ],
+  emscripten_pc_get_file__deps: ['$convertPCtoSourceLocation', 'free', '$allocateUTF8'],
   // Don't treat allocation of _emscripten_pc_get_file.ret as a leak
   emscripten_pc_get_file__noleakcheck: true,
   emscripten_pc_get_file__sig: 'pp',
@@ -3651,3 +3628,28 @@ function autoAddDeps(object, name) {
     }
   }
 }
+
+#if LEGACY_RUNTIME
+// Library function that were previosuly included as runtime functions can be
+// included by setting `LEGACY_RUNTIME`.
+DEFAULT_LIBRARY_FUNCS_TO_INCLUDE.push(
+  '$addFunction',
+  '$removeFunction',
+  '$allocate',
+  '$AsciiToString',
+  '$stringToAscii',
+  '$UTF16ToString',
+  '$stringToUTF16',
+  '$lengthBytesUTF16',
+  '$UTF32ToString',
+  '$stringToUTF32',
+  '$lengthBytesUTF32',
+  '$allocateUTF8',
+  '$allocateUTF8OnStack',
+  '$writeStringToMemory',
+  '$writeArrayToMemory',
+  '$writeAsciiToMemory',
+  '$intArrayFromString',
+  '$intArrayToString',
+);
+#endif

src/library_addfunction.js

@@ -151,11 +151,13 @@ mergeInto(LibraryManager.library, {
 
   $updateTableMap__deps: ['$getWasmTableEntry'],
   $updateTableMap: function(offset, count) {
-    for (var i = offset; i < offset + count; i++) {
-      var item = getWasmTableEntry(i);
-      // Ignore null values.
-      if (item) {
-        functionsInTableMap.set(item, i);
+    if (functionsInTableMap) {
+      for (var i = offset; i < offset + count; i++) {
+        var item = getWasmTableEntry(i);
+        // Ignore null values.
+        if (item) {
+          functionsInTableMap.set(item, i);
+        }
       }
     }
   },

src/library_bootstrap.js

@@ -23,7 +23,7 @@ mergeInto(LibraryManager.library, {
 
   // printf/puts implementations for when musl is not pulled in - very
   // partial, but enough for bootstrapping structInfo
-  printf__deps: ['$formatString'],
+  printf__deps: ['$formatString', '$intArrayToString'],
   printf__sig: 'ipp',
   printf: function(format, varargs) {
     // int printf(const char *restrict format, ...);

src/library_dylink.js

@@ -249,11 +249,7 @@ var LibraryDylink = {
   },
 
   $dlSetError__internal: true,
-  $dlSetError__deps: ['__dl_seterr',
-#if MINIMAL_RUNTIME
-   '$intArrayFromString'
-#endif
-  ],
+  $dlSetError__deps: ['__dl_seterr', '$allocateUTF8OnStack'],
   $dlSetError: function(msg) {
     withStackSave(function() {
       var cmsg = allocateUTF8OnStack(msg);
@@ -505,6 +501,7 @@ var LibraryDylink = {
     '$getDylinkMetadata', '$alignMemory', '$zeroMemory',
     '$alignMemory', '$zeroMemory',
     '$CurrentModuleWeakSymbols', '$alignMemory', '$zeroMemory',
+    '$updateTableMap',
   ],
   $loadWebAssemblyModule: function(binary, flags, handle) {
     var metadata = getDylinkMetadata(binary);

src/library_egl.js

@@ -522,9 +522,7 @@ var LibraryEGL = {
   },
 
   // EGLAPI const char * EGLAPIENTRY eglQueryString(EGLDisplay dpy, EGLint name);
-#if MINIMAL_RUNTIME
   eglQueryString__deps: ['$allocateUTF8'],
-#endif
   eglQueryString__proxy: 'sync',
   eglQueryString__sig: 'iii',
   eglQueryString: function(display, name) {

src/library_formatString.js

@@ -52,10 +52,7 @@ mergeInto(LibraryManager.library, {
   //   varargs: A pointer to the start of the arguments list.
   // Returns the resulting string string as a character array.
   $formatString__deps: ['$reallyNegative', '$convertI32PairToI53', '$convertU32PairToI53',
-                        '$reSign', '$unSign', '$strLen',
-#if MINIMAL_RUNTIME
-    , '$intArrayFromString'
-#endif
+                        '$reSign', '$unSign', '$strLen', '$intArrayFromString'
   ],
   $formatString: function(format, varargs) {
 #if ASSERTIONS

src/library_fs.js

@@ -5,7 +5,7 @@
  */
 
 mergeInto(LibraryManager.library, {
-  $FS__deps: ['$getRandomDevice', '$PATH', '$PATH_FS', '$TTY', '$MEMFS', '$asyncLoad',
+  $FS__deps: ['$getRandomDevice', '$PATH', '$PATH_FS', '$TTY', '$MEMFS', '$asyncLoad', '$intArrayFromString',
 #if LibraryManager.has('library_idbfs.js')
     '$IDBFS',
 #endif

src/library_glew.js

@@ -20,7 +20,7 @@
  */
 
 var LibraryGLEW = {
-  $GLEW__deps: ['glGetString'],
+  $GLEW__deps: ['glGetString', '$allocateUTF8'],
   $GLEW: {
     isLinaroFork: 1,
     extensions: null,

src/library_glfw.js

@@ -81,6 +81,7 @@ var LibraryGLFW = {
 
   $GLFW__deps: ['emscripten_get_now', '$GL', '$Browser', '$GLFW_Window',
     '$callUserCallback',
+    '$allocateUTF8',
 #if FILESYSTEM
     '$FS',
 #endif

src/library_openal.js

@@ -2402,6 +2402,7 @@ var LibraryOpenAL = {
 
   alcGetString__proxy: 'sync',
   alcGetString__sig: 'iii',
+  alcGetString__deps: ['$allocateUTF8'],
   alcGetString: function(deviceId, param) {
     if (AL.alcStringCache[param]) {
       return AL.alcStringCache[param];
@@ -2657,7 +2658,7 @@ var LibraryOpenAL = {
 
   emscripten_alcGetStringiSOFT__proxy: 'sync',
   emscripten_alcGetStringiSOFT__sig: 'iiii',
-  emscripten_alcGetStringiSOFT__deps: ['alcGetString'],
+  emscripten_alcGetStringiSOFT__deps: ['alcGetString', '$allocateUTF8'],
   emscripten_alcGetStringiSOFT: function(deviceId, param, index) {
     if (!(deviceId in AL.deviceRefCounts)) {
 #if OPENAL_DEBUG
@@ -3054,6 +3055,7 @@ var LibraryOpenAL = {
 
   alGetString__proxy: 'sync',
   alGetString__sig: 'ii',
+  alGetString__deps: ['$allocateUTF8'],
   alGetString: function(param) {
     if (AL.stringCache[param]) {
       return AL.stringCache[param];

src/library_sdl.js

@@ -1779,6 +1779,7 @@ var LibrarySDL = {
 
   SDL_GetKeyName__proxy: 'sync',
   SDL_GetKeyName__sig: 'ii',
+  SDL_GetKeyName__deps: ['$allocateUTF8'],
   SDL_GetKeyName: function(key) {
     if (!SDL.keyName) {
       SDL.keyName = allocateUTF8('unknown key');
@@ -1841,6 +1842,7 @@ var LibrarySDL = {
 
   SDL_GetError__proxy: 'sync',
   SDL_GetError__sig: 'i',
+  SDL_GetError__deps: ['$allocateUTF8'],
   SDL_GetError: function() {
     if (!SDL.errorMessage) {
       SDL.errorMessage = allocateUTF8("unknown SDL-emscripten error");
@@ -3573,6 +3575,7 @@ var LibrarySDL = {
 
   SDL_JoystickName__proxy: 'sync',
   SDL_JoystickName__sig: 'ii',
+  SDL_JoystickName__deps: ['$allocateUTF8'],
   SDL_JoystickName: function(deviceIndex) {
     var gamepad = SDL.getGamepad(deviceIndex);
     if (gamepad) {
@@ -3712,6 +3715,7 @@ var LibrarySDL = {
   },
 
   SDL_GetNumAudioDrivers: function() { return 1 },
+  SDL_GetCurrentAudioDriver__deps: ['$allocateUTF8'],
   SDL_GetCurrentAudioDriver: function() {
     return allocateUTF8('Emscripten Audio');
   },