Use musl implementation of gethostbyname/gethostbyaddr functions. NFC (#21067)

Instead of implementing these directly in JS, we now implement
a much lower level lookup function and rely on musl for all the higher
level stuff.

`gethostbyname` bottoms out in `_emscripten_lookup_name` and
`gethostbyaddr` bottomrs out in `getnameinfo`

Author: sbc100

src/library.js

@@ -1824,57 +1824,11 @@ addToLibrary({
     }
   },
 
-  // note: lots of leaking here!
-  gethostbyaddr__deps: ['$DNS', '$getHostByName', '$inetNtop4', '$setErrNo'],
-  gethostbyaddr__proxy: 'sync',
-  gethostbyaddr: (addr, addrlen, type) => {
-    if (type !== {{{ cDefs.AF_INET }}}) {
-      setErrNo({{{ cDefs.EAFNOSUPPORT }}});
-      // TODO: set h_errno
-      return null;
-    }
-    addr = {{{ makeGetValue('addr', '0', 'i32') }}}; // addr is in_addr
-    var host = inetNtop4(addr);
-    var lookup = DNS.lookup_addr(host);
-    if (lookup) {
-      host = lookup;
-    }
-    return getHostByName(host);
-  },
-
-  gethostbyname__deps: ['$getHostByName'],
-  gethostbyname__proxy: 'sync',
-  gethostbyname: (name) => getHostByName(UTF8ToString(name)),
-
-  $getHostByName__deps: ['malloc', '$stringToNewUTF8', '$DNS', '$inetPton4'],
-  $getHostByName: (name) => {
-    // generate hostent
-    var ret = _malloc({{{ C_STRUCTS.hostent.__size__ }}}); // XXX possibly leaked, as are others here
-    var nameBuf = stringToNewUTF8(name);
-    {{{ makeSetValue('ret', C_STRUCTS.hostent.h_name, 'nameBuf', POINTER_TYPE) }}};
-    var aliasesBuf = _malloc(4);
-    {{{ makeSetValue('aliasesBuf', '0', '0', POINTER_TYPE) }}};
-    {{{ makeSetValue('ret', C_STRUCTS.hostent.h_aliases, 'aliasesBuf', 'i8**') }}};
-    var afinet = {{{ cDefs.AF_INET }}};
-    {{{ makeSetValue('ret', C_STRUCTS.hostent.h_addrtype, 'afinet', 'i32') }}};
-    {{{ makeSetValue('ret', C_STRUCTS.hostent.h_length, '4', 'i32') }}};
-    var addrListBuf = _malloc(12);
-    {{{ makeSetValue('addrListBuf', '0', 'addrListBuf+8', POINTER_TYPE) }}};
-    {{{ makeSetValue('addrListBuf', '4', '0', POINTER_TYPE) }}};
-    {{{ makeSetValue('addrListBuf', '8', 'inetPton4(DNS.lookup_name(name))', 'i32') }}};
-    {{{ makeSetValue('ret', C_STRUCTS.hostent.h_addr_list, 'addrListBuf', 'i8**') }}};
-    return ret;
-  },
-
-  gethostbyname_r__deps: ['gethostbyname', 'memcpy', 'free'],
-  gethostbyname_r__proxy: 'sync',
-  gethostbyname_r: (name, ret, buf, buflen, out, err) => {
-    var data = _gethostbyname(name);
-    _memcpy(ret, data, {{{ C_STRUCTS.hostent.__size__ }}});
-    _free(data);
-    {{{ makeSetValue('err', '0', '0', 'i32') }}};
-    {{{ makeSetValue('out', '0', 'ret', '*') }}};
-    return 0;
+  _emscripten_lookup_name__deps: ['$UTF8ToString', '$DNS', '$inetPton4'],
+  _emscripten_lookup_name: (name) => {
+    // uint32_t _emscripten_lookup_name(const char *name);
+    var nameString = UTF8ToString(name);
+    return inetPton4(DNS.lookup_name(nameString));
   },
 
   getaddrinfo__deps: ['$Sockets', '$DNS', '$inetPton4', '$inetNtop4', '$inetPton6', '$inetNtop6', '$writeSockaddr', 'malloc', 'htonl'],

src/library_sigs.js

@@ -323,6 +323,7 @@ sigs = {
   _emscripten_fs_load_embedded_files__sig: 'vp',
   _emscripten_get_now_is_monotonic__sig: 'i',
   _emscripten_get_progname__sig: 'vpi',
+  _emscripten_lookup_name__sig: 'ip',
   _emscripten_notify_mailbox_postmessage__sig: 'vppp',
   _emscripten_push_main_loop_blocker__sig: 'vppp',
   _emscripten_push_uncounted_main_loop_blocker__sig: 'vppp',
@@ -966,9 +967,6 @@ sigs = {
   filledEllipseRGBA__sig: 'ipiiiiiiii',
   getaddrinfo__sig: 'ipppp',
   getentropy__sig: 'ipp',
-  gethostbyaddr__sig: 'ppii',
-  gethostbyname__sig: 'pp',
-  gethostbyname_r__sig: 'ipppppp',
   getnameinfo__sig: 'ipipipii',
   getprotobyname__sig: 'pp',
   getprotobynumber__sig: 'pi',

system/lib/libc/emscripten_internal.h

@@ -146,6 +146,8 @@ void __resumeException(void* exn);
 void __cxa_call_unexpected(void* exn);
 void llvm_eh_typeid_for(void* exn);
 
+uint32_t _emscripten_lookup_name(const char *name);
+
 #ifdef __cplusplus
 }
 #endif

system/lib/libc/lookup_name.c

@@ -0,0 +1,34 @@
+// Emscripten-specific version of musl/src/network/lookup_name.c
+
+#include <assert.h>
+#include <string.h>
+#include "musl/src/network/lookup.h"
+#include "emscripten_internal.h"
+
+int __lookup_name(struct address buf[static MAXADDRS], char canon[static 256], const char *name, int family, int flags)
+{
+	/* We currently only support the callsite in gethostbyname2_r which
+	 * passes AI_CANONNAME.  Remove this assertion when if we ever expand
+	 * this support. */
+	assert(flags == AI_CANONNAME);
+
+	if (family != AF_INET) {
+		return EAI_SYSTEM;
+	}
+
+	/* This hunk is a duplicated from musl/src/network/lookup_name.c */
+	*canon = 0;
+	if (name) {
+		/* reject empty name and check len so it fits into temp bufs */
+		size_t l = strnlen(name, 255);
+		if (l-1 >= 254)
+			return EAI_NONAME;
+		memcpy(canon, name, l+1);
+	}
+
+	/* We only support a single address */
+	uint32_t addr = _emscripten_lookup_name(name);
+	memset(&buf[0], 0, sizeof(buf[0]));
+	memcpy(&buf[0].addr, &addr, sizeof(addr));
+	return 1;
+}

test/other/metadce/test_metadce_hello_O0.gzsize

@@ -1 +1 @@
-8429
+8422

test/other/metadce/test_metadce_hello_O0.jssize

@@ -1 +1 @@
-23103
+23089

test/other/metadce/test_metadce_minimal_O0.gzsize

@@ -1 +1 @@
-7256
+7251

test/other/metadce/test_metadce_minimal_O0.jssize

@@ -1 +1 @@
-19727
+19713

test/other/test_support_errno.c

@@ -8,12 +8,13 @@
 #include <stdio.h>
 #include <errno.h>
 #include <string.h>
-#include <sys/types.h>
-#include <netdb.h>
+#include <emscripten/em_asm.h>
+#include <emscripten/em_js.h>
+
+EM_JS_DEPS(test, "$setErrNo");
 
 int main() {
-  void* rtn = gethostbyaddr(NULL, 0, 0);
-  printf("rtn     : %p\n", rtn);
+  EM_ASM(setErrNo(5));
   printf("errno   : %d\n", errno);
   printf("strerror: %s\n", strerror(errno));
   return 0;

test/other/test_support_errno.out

@@ -1,3 +1,2 @@
-rtn     : 0
 errno   : 5
 strerror: Address family not supported by protocol

test/other/test_support_errno_disabled.out

@@ -1,3 +1,3 @@
-rtn     : 0
+failed to set errno from JS
 errno   : 0
 strerror: No error information

test/other/test_unoptimized_code_size.js.size

@@ -1 +1 @@
-58284
+58265

test/other/test_unoptimized_code_size_strict.js.size

@@ -1 +1 @@
-57140
+57121

test/test_other.py

@@ -11918,7 +11918,7 @@ def test_default_to_cxx(self):
     'minimal': (['-sMINIMAL_RUNTIME', '-sSUPPORT_ERRNO'],),
   })
   def test_support_errno(self, args):
-    self.emcc_args += args
+    self.emcc_args += args + ['-sEXPORTED_FUNCTIONS=_main,___errno_location']
 
     self.do_other_test('test_support_errno.c')
     size_default = os.path.getsize('test_support_errno.js')

tools/system_libs.py

@@ -29,6 +29,8 @@
                 'listen.c', 'accept.c', 'getsockname.c', 'getpeername.c', 'send.c',
                 'recv.c', 'sendto.c', 'recvfrom.c', 'sendmsg.c', 'recvmsg.c',
                 'getsockopt.c', 'setsockopt.c', 'freeaddrinfo.c',
+                'gethostbyaddr.c', 'gethostbyaddr_r.c', 'gethostbyname.c',
+                'gethostbyname_r.c', 'gethostbyname2.c', 'gethostbyname2_r.c',
                 'in6addr_any.c', 'in6addr_loopback.c', 'accept4.c']
 
 # Experimental: Setting EMCC_USE_NINJA will cause system libraries to get built with ninja rather
@@ -1064,12 +1066,12 @@ def get_files(self):
     ignore += [
         'memcpy.c', 'memset.c', 'memmove.c', 'getaddrinfo.c', 'getnameinfo.c',
         'res_query.c', 'res_querydomain.c',
-        'proto.c', 'gethostbyaddr.c', 'gethostbyaddr_r.c', 'gethostbyname.c',
-        'gethostbyname2_r.c', 'gethostbyname_r.c', 'gethostbyname2.c',
+        'proto.c',
         'syscall.c', 'popen.c', 'pclose.c',
         'getgrouplist.c', 'initgroups.c', 'wordexp.c', 'timer_create.c',
         'getentropy.c',
         'getauxval.c',
+        'lookup_name.c',
         # 'process' exclusion
         'fork.c', 'vfork.c', 'posix_spawn.c', 'posix_spawnp.c', 'execve.c', 'waitid.c', 'system.c',
         '_Fork.c',
@@ -1258,6 +1260,7 @@ def get_files(self):
           'mktime.c',
           'tzset.c',
           'kill.c',
+          'lookup_name.c',
           'pthread_sigmask.c',
           'raise.c',
           'sigaction.c',