Add UDP support to dart:io

Add UDP support to dart:io

Initial implementation of UDP support in dart:io

Added a RawDatagramSocket which gets the same events as a RawSocket
except that READ_CLOSED is never received.

The receive from a RawDatagramSocket returns a Datagram object where
the senders address and port is presten together with the data.

The send on a RawDatagramSocket takes data and the address and port of
the destination.

Extended the number of socket options to support UDP.

Added getting the socket options.

R=ajohnsen@google.com
BUG=http://dartbug.com/1975

Committed: https://code.google.com/p/dart/source/detail?r=31098

[Søren Gjesse, 2013-11-25 16:35:47]

Linux implementation only for now.

We could consider sharing some of the code between RawSocket and
RawDatagramSocket in a common superclass.

[Anders Johnsen, 2013-11-25 18:00:06]

First round of interface comments. Yay for UDP! :)

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart
File sdk/lib/io/socket.dart (right):

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:321: static const SocketOption _IP_MULTICAST_LOOP = const
SocketOption._(1);
Should we move private ones to start by, say 16?

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:528: * as there is no connection which can be closed by a
remote peer.
'[...] as an UDP socket cannot be closed by a remote peer.' ?

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:530: abstract class RawDatagramSocket extends Stream {
Can the Stream type be Stream<RawSocketEvent> ?

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:535: factory RawDatagramSocket(InternetAddress address,
int port) =>
I think this should be like a Socket and ServerSocket, returning a Future. If
not, then one could argue that ServerSocket.bind should be a factory as well?

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:551: void close();
Does directional-close work for UDP sockets? (I'm thinking about shutdown)

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:565: bool sendTo(List<data> buffer, InternetAddress
address, int port);
Having both send and sendTo, with so little difference, seems a bit redundant.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:579: void join(InternetAddress group, [NetworkInterface
interface]);
This is really cool, with the optional interface. Could the interface be a named
argument?

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:587: void leave(InternetAddress group, [NetworkInterface
interface]);
Ditto.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:601: bool writeEventsEnabled;
Is it true by default? I thought it is only set to true, when one cannot send a
message.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:604: * Set or get, whether multicast traffic is looped
back to the host.
This, and those below, needs a default value in documentation.

[Søren Gjesse, 2013-11-28 13:58:52]

ptal

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart
File sdk/lib/io/socket.dart (right):

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:321: static const SocketOption _IP_MULTICAST_LOOP = const
SocketOption._(1);
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Should we move private ones to start by, say 16?

The numbers are not visible outside, so it does not really matter.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:528: * as there is no connection which can be closed by a
remote peer.
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> '[...] as an UDP socket cannot be closed by a remote peer.' ?

Done.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:530: abstract class RawDatagramSocket extends Stream {
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Can the Stream type be Stream<RawSocketEvent> ?

Done.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:535: factory RawDatagramSocket(InternetAddress address,
int port) =>
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> I think this should be like a Socket and ServerSocket, returning a Future. If
> not, then one could argue that ServerSocket.bind should be a factory as well?

I will make it return a future, and change the type from InternetAddress to
dynamic to support both InternetAddress and hostname.

My initial thoughts was that for most UDP applications one would use a numeric
InternetAddress, and therefore a future would not be needed as no DNS lookup
could be involved.

Lets keep this consistent, we can then consider normal constructors for
RawServerSocket, RawSocket and this with InternetAddress. But that might just
cause confusion.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:551: void close();
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Does directional-close work for UDP sockets? (I'm thinking about shutdown)

No the shutdown is a TCP thing.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:565: bool sendTo(List<data> buffer, InternetAddress
address, int port);
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Having both send and sendTo, with so little difference, seems a bit redundant.

Removed one.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:579: void join(InternetAddress group, [NetworkInterface
interface]);
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> This is really cool, with the optional interface. Could the interface be a
named
> argument?

Done.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:587: void leave(InternetAddress group, [NetworkInterface
interface]);
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Ditto.

Done.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:601: bool writeEventsEnabled;
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> Is it true by default? I thought it is only set to true, when one cannot send
a
> message.

It starts out as true.

https://codereview.chromium.org/85993002/diff/1/sdk/lib/io/socket.dart#newcod...
sdk/lib/io/socket.dart:604: * Set or get, whether multicast traffic is looped
back to the host.
On 2013/11/25 18:00:06, Anders Johnsen wrote:
> This, and those below, needs a default value in documentation.

Done.

[Anders Johnsen, 2013-11-29 12:36:54]

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc#new...
runtime/bin/socket.cc:136: OSError error;
Unused?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc#new...
runtime/bin/socket.cc:265: if (drop_datagrams) {
Drop earlier?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h
File runtime/bin/socket.h (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h#newc...
runtime/bin/socket.h:170: static int SendTo(intptr_t fd, const void* buffer,
intptr_t num_bytes,
All on one line or each on its own?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h#newc...
runtime/bin/socket.h:172: static int RecvFrom(intptr_t fd, void* buffer,
intptr_t num_bytes,
Ditto

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux.cc
File runtime/bin/socket_linux.cc (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:30: if (TEMP_FAILURE_RETRY(getnameinfo(sa,
Use FormatNumericAddress here?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:130: recvfrom(fd, buffer, num_bytes, 0, &addr->addr,
&addr_len);
TEMP_FAILURE_RETRY?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:520: ok = TEMP_FAILURE_RETRY(getsockopt(fd,
Set ipv4/ipv6 values above and only have one call? (likewise below).

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
File runtime/bin/socket_patch.dart (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:128: 
intentional?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:486: reportError(result, "Receive failed");
Put in scheduleMicrotask.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:528: var result = nativeSendTo(
Use result?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:759: bool getOption(SocketOption option) {
I suppose the result is dynamic?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1378: _RawDatagramSocket(this._socket) {
Update with zone wrappers, like in RawSocket.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1402: static Future<RawDatagramSocket> bind(
patch method?

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1428: assert(result == buffer.length);
It can be 0?

https://codereview.chromium.org/85993002/diff/20001/tests/standalone/io/raw_d...
File tests/standalone/io/raw_datagram_socket_test.dart (right):

https://codereview.chromium.org/85993002/diff/20001/tests/standalone/io/raw_d...
tests/standalone/io/raw_datagram_socket_test.dart:203:
sender.send(createDataPackage(seq), bindAddress, receiver.port);
Check send return value.

[Søren Gjesse, 2013-12-12 11:38:46]

PTAL

Implementation for all Platforms.

The Android code has been verified to compile, and we don't currently run the IO
tests on Android. Testing on Mac OS is disabled, as I have not been able to get
multicast to work on my laptop yet.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc#new...
runtime/bin/socket.cc:136: OSError error;
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Unused?

Yes, removed.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.cc#new...
runtime/bin/socket.cc:265: if (drop_datagrams) {
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Drop earlier?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h
File runtime/bin/socket.h (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h#newc...
runtime/bin/socket.h:170: static int SendTo(intptr_t fd, const void* buffer,
intptr_t num_bytes,
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> All on one line or each on its own?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket.h#newc...
runtime/bin/socket.h:172: static int RecvFrom(intptr_t fd, void* buffer,
intptr_t num_bytes,
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Ditto

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux.cc
File runtime/bin/socket_linux.cc (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:30: if (TEMP_FAILURE_RETRY(getnameinfo(sa,
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Use FormatNumericAddress here?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:130: recvfrom(fd, buffer, num_bytes, 0, &addr->addr,
&addr_len);
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> TEMP_FAILURE_RETRY?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_linux....
runtime/bin/socket_linux.cc:520: ok = TEMP_FAILURE_RETRY(getsockopt(fd,
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Set ipv4/ipv6 values above and only have one call? (likewise below).

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
File runtime/bin/socket_patch.dart (right):

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:128: 
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> intentional? 

No, reverted.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:486: reportError(result, "Receive failed");
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Put in scheduleMicrotask.

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:528: var result = nativeSendTo(
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Use result?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:759: bool getOption(SocketOption option) {
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> I suppose the result is dynamic?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1378: _RawDatagramSocket(this._socket) {
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Update with zone wrappers, like in RawSocket.

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1402: static Future<RawDatagramSocket> bind(
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> patch method?

Done.

https://codereview.chromium.org/85993002/diff/20001/runtime/bin/socket_patch....
runtime/bin/socket_patch.dart:1428: assert(result == buffer.length);
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> It can be 0?

Yes, Changed to just call the native socket as the error handling is there.

The return type is now int and returns the number of bytes written.

https://codereview.chromium.org/85993002/diff/20001/tests/standalone/io/raw_d...
File tests/standalone/io/raw_datagram_socket_test.dart (right):

https://codereview.chromium.org/85993002/diff/20001/tests/standalone/io/raw_d...
tests/standalone/io/raw_datagram_socket_test.dart:203:
sender.send(createDataPackage(seq), bindAddress, receiver.port);
On 2013/11/29 12:36:54, Anders Johnsen wrote:
> Check send return value.

Done.

[Anders Johnsen, 2013-12-12 12:19:58]

LGTM, yay, this is very cool! :) Adding isMulticastEnabled may help on mac (I
read that it's disabled by default on devices).

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
File runtime/bin/eventhandler_win.cc (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:39: if (operation == kRecvFrom) {
Move this to AllocateRecvFromBuffer?

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:40: actual_size += sizeof(socklen_t) +
sizeof(struct sockaddr_storage);
Make the size a constant in OverlappedBuffer.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:581: num_bytes = data_ready_->Read(buffer,
num_bytes);
Will this allow us to read only a part of a UDP packet?

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:620: if (!IssueSendTo(sa, sa_len)) return -1;
Can this leak the pending_write?

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:1278: OverlappedBuffer* buffer =
Unused?

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
File runtime/bin/eventhandler_win.h (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:13: #include <ws2tcpip.h>
This looks odd - importing a TCP/IP header for UDP CL? Is it just how Windows
headers was made? :)

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:104: static int additional_size =
static -> const?

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:105: sizeof(struct sockaddr_storage) +
sizeof(socklen_t);
Indentation

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc
File runtime/bin/socket.cc (left):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ol...
runtime/bin/socket.cc:28: 
Re-add this.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ne...
runtime/bin/socket.cc:122: 
Double-space, and below.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ne...
runtime/bin/socket.cc:127: Dart_GetNativeArgument(args, 2),
If you extract arg 2 and 3 above, it'll be easier to read this, and take up
fewer lines.

https://codereview.chromium.org/85993002/diff/190001/sdk/lib/io/socket.dart
File sdk/lib/io/socket.dart (right):

https://codereview.chromium.org/85993002/diff/190001/sdk/lib/io/socket.dart#n...
sdk/lib/io/socket.dart:536: abstract class RawDatagramSocket extends
Stream<RawSocketEvent> {
Maybe add a static bool 'isMulticastEnabled(NetworkInterface interface)'?

[Søren Gjesse, 2013-12-12 15:44:38]

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
File runtime/bin/eventhandler_win.cc (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:39: if (operation == kRecvFrom) {
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Move this to AllocateRecvFromBuffer?

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:40: actual_size += sizeof(socklen_t) +
sizeof(struct sockaddr_storage);
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Make the size a constant in OverlappedBuffer.

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:581: num_bytes = data_ready_->Read(buffer,
num_bytes);
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Will this allow us to read only a part of a UDP packet?

It will, I changed it to dispose of the rest, so that the next call to RecvFrom
should always get the next datagram or nothing. This is how recvfrom on Linux
and Mac OS works if the buffer is smaller than the datagram.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:620: if (!IssueSendTo(sa, sa_len)) return -1;
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Can this leak the pending_write?

No, IssueSendTo will dispose of it before returning false (same for Write
above).

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.cc:1278: OverlappedBuffer* buffer =
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Unused?

Yes, removed.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
File runtime/bin/eventhandler_win.h (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:13: #include <ws2tcpip.h>
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> This looks odd - importing a TCP/IP header for UDP CL? Is it just how Windows
> headers was made? :)

It contains the typedef for socklen_t.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:104: static int additional_size =
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> static -> const?

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/eventhandler...
runtime/bin/eventhandler_win.h:105: sizeof(struct sockaddr_storage) +
sizeof(socklen_t);
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Indentation

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc
File runtime/bin/socket.cc (left):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ol...
runtime/bin/socket.cc:28: 
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Re-add this.

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc
File runtime/bin/socket.cc (right):

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ne...
runtime/bin/socket.cc:122: 
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Double-space, and below.

Done.

https://codereview.chromium.org/85993002/diff/190001/runtime/bin/socket.cc#ne...
runtime/bin/socket.cc:127: Dart_GetNativeArgument(args, 2),
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> If you extract arg 2 and 3 above, it'll be easier to read this, and take up
> fewer lines.

Done.

https://codereview.chromium.org/85993002/diff/190001/sdk/lib/io/socket.dart
File sdk/lib/io/socket.dart (right):

https://codereview.chromium.org/85993002/diff/190001/sdk/lib/io/socket.dart#n...
sdk/lib/io/socket.dart:536: abstract class RawDatagramSocket extends
Stream<RawSocketEvent> {
On 2013/12/12 12:19:58, Anders Johnsen wrote:
> Maybe add a static bool 'isMulticastEnabled(NetworkInterface interface)'?

We should definitely add more information like this.

[Søren Gjesse, 2013-12-12 15:45:41]

Committed patchset #13 manually as r31098 (presubmit successful).