Move PROXY_TO_PTHREAD handling into native code. NFC (#17153)

Fixes: #17144

Author: sbc100

emcc.py

@@ -1741,7 +1741,10 @@ def phase_linker_setup(options, state, newargs, user_settings):
         settings.EXPECT_MAIN = 0
     else:
       assert not settings.EXPORTED_FUNCTIONS
-      settings.EXPORTED_FUNCTIONS = ['_main']
+      # With PROXY_TO_PTHREAD we don't export `main` at all but instead
+      # we export `emscripten_proxy_main` (elsewhwere).
+      if not settings.PROXY_TO_PTHREAD:
+        settings.EXPORTED_FUNCTIONS = ['_main']
 
   if settings.STANDALONE_WASM:
     # In STANDALONE_WASM mode we either build a command or a reactor.
@@ -2519,12 +2522,6 @@ def check_memory_setting(setting):
       # they are indirect calls, since that is what they do - we can't see their
       # targets statically.
       settings.ASYNCIFY_IMPORTS += ['invoke_*']
-    # with pthreads we may call main through the __call_main mechanism, which can
-    # therefore reach anything in the program, so mark it as possibly causing a
-    # sleep (the asyncify analysis doesn't look through JS, just wasm, so it can't
-    # see what it itself calls)
-    if settings.USE_PTHREADS:
-      settings.ASYNCIFY_IMPORTS += ['__call_main']
     # add the default imports
     settings.ASYNCIFY_IMPORTS += DEFAULT_ASYNCIFY_IMPORTS
 

emscripten.py

@@ -147,7 +147,7 @@ def update_settings_glue(wasm_file, metadata):
   if settings.MEMORY64:
     assert '--enable-memory64' in settings.BINARYEN_FEATURES
 
-  settings.HAS_MAIN = bool(settings.MAIN_MODULE) or settings.STANDALONE_WASM or 'main' in settings.WASM_EXPORTS or '__main_argc_argv' in settings.WASM_EXPORTS
+  settings.HAS_MAIN = bool(settings.MAIN_MODULE) or settings.PROXY_TO_PTHREAD or settings.STANDALONE_WASM or 'main' in settings.WASM_EXPORTS or '__main_argc_argv' in settings.WASM_EXPORTS
 
   # When using dynamic linking the main function might be in a side module.
   # To be safe assume they do take input parametes.
@@ -220,12 +220,6 @@ def report_missing_symbols(js_library_funcs):
     # In this mode we never expect _main in the export list.
     return
 
-  # PROXY_TO_PTHREAD only makes sense with a main(), so error if one is
-  # missing. note that when main() might arrive from another module we cannot
-  # error here.
-  if settings.PROXY_TO_PTHREAD and not settings.HAS_MAIN:
-    exit_with_error('PROXY_TO_PTHREAD proxies main() for you, but no main exists')
-
   if settings.IGNORE_MISSING_MAIN:
     # The default mode for emscripten is to ignore the missing main function allowing
     # maximum compatibility.

src/jsifier.js

@@ -17,6 +17,9 @@ global.proxiedFunctionTable = ['null'/* Reserve index 0 for an undefined functio
 
 // Mangles the given C/JS side function name to assembly level function name (adds an underscore)
 function mangleCSymbolName(f) {
+  if (f === '__main_argc_argv') {
+    f = 'main';
+  }
   return f[0] == '$' ? f.substr(1) : '_' + f;
 }
 
@@ -53,6 +56,9 @@ function isDefined(symName) {
   if (WASM_EXPORTS.has(symName) || SIDE_MODULE_EXPORTS.has(symName)) {
     return true;
   }
+  if (symName == '__main_argc_argv' && SIDE_MODULE_EXPORTS.has('main')) {
+    return true;
+  }
   // 'invoke_' symbols are created at runtime in libary_dylink.py so can
   // always be considered as defined.
   if (RELOCATABLE && symName.startsWith('invoke_')) {
@@ -208,9 +214,13 @@ function ${name}(${args}) {
         }
         const isWeakImport = WEAK_IMPORTS.has(ident);
         if (!isDefined(ident) && !isWeakImport) {
+          if (PROXY_TO_PTHREAD && !MAIN_MODULE && ident == '__main_argc_argv') {
+            error('PROXY_TO_PTHREAD proxies main() for you, but no main exists');
+            return;
+          }
           let undefinedSym = ident;
           if (ident === '__main_argc_argv') {
-            undefinedSym = 'main';
+            undefinedSym = 'main/__main_argc_argv';
           }
           let msg = 'undefined symbol: ' + undefinedSym;
           if (dependent) msg += ` (referenced by ${dependent})`;
@@ -224,7 +234,7 @@ function ${name}(${args}) {
           } else if (VERBOSE || WARN_ON_UNDEFINED_SYMBOLS) {
             warn(msg);
           }
-          if (undefinedSym === 'main' && STANDALONE_WASM) {
+          if (ident === '__main_argc_argv' && STANDALONE_WASM) {
             warn('To build in STANDALONE_WASM mode without a main(), use emcc --no-entry');
           }
         }

src/library_dylink.js

@@ -475,16 +475,17 @@ var LibraryDylink = {
 
       // Export native export on the Module object.
       // TODO(sbc): Do all users want this?  Should we skip this by default?
-#if !hasExportedFunction('_main')
-      // If the main module doesn't define main it could be defined in one of
-      // the side module, and we need to handle the mangled named.
-      var module_sym = asmjsMangle(sym == '__main_argc_argv' ? 'main' : sym);
-#else
       var module_sym = asmjsMangle(sym);
-#endif
       if (!Module.hasOwnProperty(module_sym)) {
         Module[module_sym] = exports[sym];
       }
+#if !hasExportedFunction('_main')
+      // If the main module doesn't define main it could be defined in one of
+      // the side modules, and we need to handle the mangled named.
+      if (sym == '__main_argc_argv') {
+        Module['_main'] = exports[sym];
+      }
+#endif
     }
   },
 

src/library_pthread.js

@@ -807,30 +807,6 @@ var LibraryPThread = {
     return 0;
   },
 
-#if PROXY_TO_PTHREAD
-  __call_main__deps: ['exit', '$exitOnMainThread'],
-  __call_main: function(argc, argv) {
-#if !EXIT_RUNTIME
-    // EXIT_RUNTIME==0 set, keeping main thread alive by default.
-    noExitRuntime = true;
-#endif
-    var returnCode = {{{ exportedAsmFunc('_main') }}}(argc, argv);
-#if EXIT_RUNTIME
-    if (!keepRuntimeAlive()) {
-      // exitRuntime enabled, proxied main() finished in a pthread, shut down the process.
-#if PTHREADS_DEBUG
-      err('Proxied main thread finished with return code ' + returnCode + '. EXIT_RUNTIME=1 set, quitting process.');
-#endif
-      exitOnMainThread(returnCode);
-    }
-#else
-#if PTHREADS_DEBUG
-    err('Proxied main thread finished with return code ' + returnCode + '. EXIT_RUNTIME=0 set, so keeping main thread alive for asynchronous event operations.');
-#endif
-#endif
-  },
-#endif
-
   // This function is call by a pthread to signal that exit() was called and
   // that the entire process should exit.
   // This function is always called from a pthread, but is executed on the

system/lib/pthread/emscripten_proxy_main.c

@@ -6,20 +6,30 @@
  */
 
 #include <pthread.h>
+#include <stdlib.h>
 
 #include <emscripten.h>
 #include <emscripten/stack.h>
 #include <emscripten/threading.h>
+#include <emscripten/eventloop.h>
 
 static int _main_argc;
 static char** _main_argv;
 
-extern void __call_main(int argc, char** argv);
+int __main_argc_argv(int argc, char *argv[]);
+
+weak int __main_void(void) {
+  return __main_argc_argv(_main_argc, _main_argv);
+}
 
 static void* _main_thread(void* param) {
   // This is the main runtime thread for the application.
   emscripten_set_thread_name(pthread_self(), "Application main thread");
-  __call_main(_main_argc, _main_argv);
+  // Will either call user's __main_void or weak version above.
+  int rtn = __main_void();
+  if (!emscripten_runtime_keepalive_check()) {
+    exit(rtn);
+  }
   return NULL;
 }
 

tests/other/metadce/test_metadce_minimal_pthreads.exports

@@ -2,7 +2,6 @@ A
 B
 C
 D
-E
 o
 p
 q

tests/other/metadce/test_metadce_minimal_pthreads.funcs

@@ -46,7 +46,6 @@ $emscripten_stack_set_limits
 $get_tasks_for_thread
 $init_file_lock
 $init_mparams
-$main
 $memset
 $pthread_mutexattr_destroy
 $sbrk

tests/other/metadce/test_metadce_minimal_pthreads.jssize

@@ -1 +1 @@
-16131
+16048

tests/other/metadce/test_metadce_minimal_pthreads.size

@@ -1 +1 @@
-18331
+18318