Bug 1110938 - Add timeout for CPOWs (default to 0) (r=dvander)

browser/app/profile/firefox.js

@@ -1849,3 +1849,6 @@ pref("extensions.interposition.enabled", true);
 #endif
 
 pref("browser.defaultbrowser.notificationbar", false);
+
+// How many milliseconds to wait for a CPOW response from the child process.
+pref("dom.ipc.cpow.timeout", 0);

dom/ipc/ContentParent.cpp

@@ -2011,6 +2011,9 @@ ContentParent::ContentParent(mozIApplication* aApp,
                  true  /* Send registered chrome */);
 
     ContentProcessManager::GetSingleton()->AddContentProcess(this);
+
+    // Set a reply timeout for CPOWs.
+    SetReplyTimeoutMs(Preferences::GetInt("dom.ipc.cpow.timeout", 0));
 }
 
 #ifdef MOZ_NUWA_PROCESS

dom/plugins/ipc/PluginModuleParent.cpp

@@ -571,6 +571,7 @@ PluginModuleChromeParent::ShouldContinueFromReplyTimeout()
     FinishHangUI();
 #endif // XP_WIN
     TerminateChildProcess(MessageLoop::current());
+    GetIPCChannel()->CloseWithTimeout();
     return false;
 }
 

ipc/glue/MessageChannel.cpp

@@ -292,6 +292,8 @@ MessageChannel::MessageChannel(MessageListener *aListener)
     mDispatchingSyncMessage(false),
     mDispatchingSyncMessagePriority(0),
     mCurrentTransaction(0),
+    mTimedOutMessageSeqno(0),
+    mRecvdErrors(0),
     mRemoteStackDepthGuess(false),
     mSawInterruptOutMsg(false),
     mAbortOnError(false),
@@ -611,8 +613,30 @@ MessageChannel::OnMessageReceivedFromLink(const Message& aMsg)
 
     // Regardless of the Interrupt stack, if we're awaiting a sync reply,
     // we know that it needs to be immediately handled to unblock us.
-    if (AwaitingSyncReply() && aMsg.is_sync() && aMsg.is_reply()) {
+    if (aMsg.is_sync() && aMsg.is_reply()) {
+        if (aMsg.seqno() == mTimedOutMessageSeqno) {
+            // Drop the message, but allow future sync messages to be sent.
+            mTimedOutMessageSeqno = 0;
+            return;
+        }
+
+        MOZ_ASSERT(AwaitingSyncReply());
         MOZ_ASSERT(!mRecvd);
+
+        // Rather than storing errors in mRecvd, we mark them in
+        // mRecvdErrors. We need a counter because multiple replies can arrive
+        // when a timeout happens, as in the following example. Imagine the
+        // child is running slowly. The parent sends a sync message P1. It times
+        // out. The child eventually sends a sync message C1. While waiting for
+        // the C1 response, the child dispatches P1. In doing so, it sends sync
+        // message C2. At that point, it's valid for the parent to send error
+        // responses for both C1 and C2.
+        if (aMsg.is_reply_error()) {
+            mRecvdErrors++;
+            NotifyWorkerThread();
+            return;
+        }
+
         mRecvd = new Message(aMsg);
         NotifyWorkerThread();
         return;
@@ -698,6 +722,14 @@ MessageChannel::Send(Message* aMsg, Message* aReply)
 
     MonitorAutoLock lock(*mMonitor);
 
+    if (mTimedOutMessageSeqno) {
+        // Don't bother sending another sync message if a previous one timed out
+        // and we haven't received a reply for it. Once the original timed-out
+        // message receives a reply, we'll be able to send more sync messages
+        // again.
+        return false;
+    }
+
     IPC_ASSERT(aMsg->is_sync(), "can only Send() sync messages here");
     IPC_ASSERT(aMsg->priority() >= DispatchingSyncMessagePriority(),
                "can't send sync message of a lesser priority than what's being dispatched");
@@ -713,12 +745,12 @@ MessageChannel::Send(Message* aMsg, Message* aReply)
 
     msg->set_seqno(NextSeqno());
 
-    DebugOnly<int32_t> replySeqno = msg->seqno();
+    int32_t seqno = msg->seqno();
     DebugOnly<msgid_t> replyType = msg->type() + 1;
 
     AutoSetValue<bool> replies(mAwaitingSyncReply, true);
     AutoSetValue<int> prio(mAwaitingSyncReplyPriority, msg->priority());
-    AutoEnterTransaction transact(this, msg->seqno());
+    AutoEnterTransaction transact(this, seqno);
 
     int32_t transaction = mCurrentTransaction;
     msg->set_transaction_id(transaction);
@@ -750,16 +782,16 @@ MessageChannel::Send(Message* aMsg, Message* aReply)
         }
 
         // See if we've received a reply.
+        if (mRecvdErrors) {
+            mRecvdErrors--;
+            return false;
+        }
+
         if (mRecvd) {
             MOZ_ASSERT(mRecvd->is_reply(), "expected reply");
-
+            MOZ_ASSERT(!mRecvd->is_reply_error());
-            if (mRecvd->is_reply_error()) {
-                mRecvd = nullptr;
-                return false;
-            }
-
             MOZ_ASSERT(mRecvd->type() == replyType, "wrong reply type");
-            MOZ_ASSERT(mRecvd->seqno() == replySeqno);
+            MOZ_ASSERT(mRecvd->seqno() == seqno);
             MOZ_ASSERT(mRecvd->is_sync());
 
             *aReply = Move(*mRecvd);
@@ -767,6 +799,8 @@ MessageChannel::Send(Message* aMsg, Message* aReply)
             return true;
         }
 
+        MOZ_ASSERT(!mTimedOutMessageSeqno);
+
         bool maybeTimedOut = !WaitForSyncNotify();
 
         if (!Connected()) {
@@ -774,8 +808,13 @@ MessageChannel::Send(Message* aMsg, Message* aReply)
             return false;
         }
 
-        if (maybeTimedOut && !ShouldContinueFromTimeout())
+        // We only time out a message if it initiated a new transaction (i.e.,
+        // if neither side has any other message Sends on the stack).
+        bool canTimeOut = transaction == seqno;
+        if (maybeTimedOut && canTimeOut && !ShouldContinueFromTimeout()) {
+            mTimedOutMessageSeqno = seqno;
             return false;
+        }
     }
 
     return true;
@@ -1089,7 +1128,14 @@ MessageChannel::DispatchSyncMessage(const Message& aMsg)
     bool& blockingVar = ShouldBlockScripts() ? gParentIsBlocked : dummy;
 
     Result rv;
-    {
+    if (mTimedOutMessageSeqno) {
+        // If the other side sends a message in response to one of our messages
+        // that we've timed out, then we reply with an error.
+        //
+        // We even reject messages that were sent before the other side even got
+        // to our timed out message.
+        rv = MsgNotAllowed;
+    } else {
         AutoSetValue<bool> blocked(blockingVar, true);
         AutoSetValue<bool> sync(mDispatchingSyncMessage, true);
         AutoSetValue<int> prioSet(mDispatchingSyncMessagePriority, prio);
@@ -1104,6 +1150,7 @@ MessageChannel::DispatchSyncMessage(const Message& aMsg)
         reply->set_reply_error();
     }
     reply->set_seqno(aMsg.seqno());
+    reply->set_transaction_id(aMsg.transaction_id());
 
     MonitorAutoLock lock(*mMonitor);
     if (ChannelConnected == mChannelState) {
@@ -1361,21 +1408,6 @@ MessageChannel::ShouldContinueFromTimeout()
         return true;
     }
 
-    if (!cont) {
-        // NB: there's a sublety here.  If parents were allowed to send sync
-        // messages to children, then it would be possible for this
-        // synchronous close-on-timeout to race with async |OnMessageReceived|
-        // tasks arriving from the child, posted to the worker thread's event
-        // loop.  This would complicate cleanup of the *Channel.  But since
-        // IPDL forbids this (and since it doesn't support children timing out
-        // on parents), the parent can only block on interrupt messages to the child,
-        // and in that case arriving async messages are enqueued to the interrupt 
-        // channel's special queue.  They're then ignored because the channel
-        // state changes to ChannelTimeout (i.e. !Connected).
-        SynchronouslyClose();
-        mChannelState = ChannelTimeout;
-    }
-
     return cont;
 }
 
@@ -1618,6 +1650,19 @@ MessageChannel::CloseWithError()
     PostErrorNotifyTask();
 }
 
+void
+MessageChannel::CloseWithTimeout()
+{
+    AssertWorkerThread();
+
+    MonitorAutoLock lock(*mMonitor);
+    if (ChannelConnected != mChannelState) {
+        return;
+    }
+    SynchronouslyClose();
+    mChannelState = ChannelTimeout;
+}
+
 void
 MessageChannel::BlockScripts()
 {

ipc/glue/MessageChannel.h

@@ -84,6 +84,8 @@ class MessageChannel : HasResultCodes
     // for process links only, not thread links.
     void CloseWithError();
 
+    void CloseWithTimeout();
+
     void SetAbortOnError(bool abort)
     {
         mAbortOnError = true;
@@ -518,10 +520,29 @@ class MessageChannel : HasResultCodes
        int32_t mOldTransaction;
     };
 
+    // If a sync message times out, we store its sequence number here. Any
+    // future sync messages will fail immediately. Once the reply for original
+    // sync message is received, we allow sync messages again.
+    //
+    // When a message times out, nothing is done to inform the other side. The
+    // other side will eventually dispatch the message and send a reply. Our
+    // side is responsible for replying to all sync messages sent by the other
+    // side when it dispatches the timed out message. The response is always an
+    // error.
+    //
+    // A message is only timed out if it initiated a transaction. This avoids
+    // hitting a lot of corner cases with message nesting that we don't really
+    // care about.
+    int32_t mTimedOutMessageSeqno;
+
     // If waiting for the reply to a sync out-message, it will be saved here
     // on the I/O thread and then read and cleared by the worker thread.
     nsAutoPtr<Message> mRecvd;
 
+    // If a sync message reply that is an error arrives, we increment this
+    // counter rather than storing it in mRecvd.
+    size_t mRecvdErrors;
+
     // Queue of all incoming messages, except for replies to sync and urgent
     // messages, which are delivered directly to mRecvd, and any pending urgent
     // incall, which is stored in mPendingUrgentRequest.

ipc/ipdl/test/cxx/PTestUrgentHangs.ipdl

@@ -0,0 +1,24 @@
+namespace mozilla {
+namespace _ipdltest {
+
+prio(normal upto high) sync protocol PTestUrgentHangs
+{
+parent:
+    prio(high) sync Test1_2();
+
+    prio(high) sync TestInner();
+
+child:
+    prio(high) sync Test1_1();
+    prio(high) sync Test1_3();
+
+    prio(high) sync Test2();
+
+    prio(high) sync Test3();
+
+    async Test4();
+    prio(high) sync Test4_1();
+};
+
+} // namespace _ipdltest
+} // namespace mozilla

ipc/ipdl/test/cxx/TestUrgentHangs.cpp

@@ -0,0 +1,170 @@
+/* -*- Mode: C++; tab-width: 4; indent-tabs-mode: nil; c-basic-offset: 4 -*-
+ * vim: sw=4 ts=4 et :
+ */
+#include "TestUrgentHangs.h"
+
+#include "IPDLUnitTests.h"      // fail etc.
+#if defined(OS_POSIX)
+#include <unistd.h>
+#else
+#include <windows.h>
+#endif
+
+template<>
+struct RunnableMethodTraits<mozilla::_ipdltest::TestUrgentHangsParent>
+{
+    static void RetainCallee(mozilla::_ipdltest::TestUrgentHangsParent* obj) { }
+    static void ReleaseCallee(mozilla::_ipdltest::TestUrgentHangsParent* obj) { }
+};
+
+namespace mozilla {
+namespace _ipdltest {
+
+//-----------------------------------------------------------------------------
+// parent
+
+TestUrgentHangsParent::TestUrgentHangsParent()
+{
+    MOZ_COUNT_CTOR(TestUrgentHangsParent);
+}
+
+TestUrgentHangsParent::~TestUrgentHangsParent()
+{
+    MOZ_COUNT_DTOR(TestUrgentHangsParent);
+}
+
+void
+TestUrgentHangsParent::Main()
+{
+    SetReplyTimeoutMs(1000);
+
+    // Should succeed despite the nested sleep call because the content process
+    // responded to the transaction.
+    if (!SendTest1_1())
+        fail("sending Test1_1");
+
+    // Fails with a timeout.
+    if (SendTest2())
+        fail("sending Test2");
+
+    // Also fails since we haven't gotten a response for Test2 yet.
+    if (SendTest3())
+        fail("sending Test3");
+
+    // Do a second round of testing once the reply to Test2 comes back.
+    MessageLoop::current()->PostDelayedTask(
+        FROM_HERE,
+        NewRunnableMethod(this, &TestUrgentHangsParent::FinishTesting),
+        3000);
+}
+
+void
+TestUrgentHangsParent::FinishTesting()
+{
+    // Send an async message that waits 2 seconds and then sends a sync message
+    // (which should be processed).
+    if (!SendTest4())
+        fail("sending Test4");
+
+    // Send a sync message that will time out because the child is waiting
+    // inside RecvTest4.
+    if (SendTest4_1())
+        fail("sending Test4_1");
+
+    // Close the channel after the child finishes its work in RecvTest4.
+    MessageLoop::current()->PostDelayedTask(
+        FROM_HERE,
+        NewRunnableMethod(this, &TestUrgentHangsParent::Close),
+        3000);
+}
+
+bool
+TestUrgentHangsParent::RecvTest1_2()
+{
+    if (!SendTest1_3())
+        fail("sending Test1_3");
+    return true;
+}
+
+bool
+TestUrgentHangsParent::RecvTestInner()
+{
+    fail("TestInner should never be dispatched");
+    return true;
+}
+
+//-----------------------------------------------------------------------------
+// child
+
+bool
+TestUrgentHangsChild::RecvTest1_1()
+{
+    if (!SendTest1_2())
+        fail("sending Test1_2");
+
+    return true;
+}
+
+bool
+TestUrgentHangsChild::RecvTest1_3()
+{
+    sleep(2);
+
+    return true;
+}
+
+bool
+TestUrgentHangsChild::RecvTest2()
+{
+    sleep(2);
+
+    // Should fail because of the timeout.
+    if (SendTestInner())
+        fail("sending TestInner");
+
+    return true;
+}
+
+bool
+TestUrgentHangsChild::RecvTest3()
+{
+    fail("RecvTest3 should never be called");
+    return true;
+}
+
+bool
+TestUrgentHangsChild::RecvTest4()
+{
+    sleep(2);
+
+    // This should fail because Test4_1 timed out and hasn't gotten a response
+    // yet.
+    if (SendTestInner())
+        fail("sending TestInner");
+
+    return true;
+}
+
+bool
+TestUrgentHangsChild::RecvTest4_1()
+{
+    // This should fail because Test4_1 timed out and hasn't gotten a response
+    // yet.
+    if (SendTestInner())
+        fail("sending TestInner");
+
+    return true;
+}
+
+TestUrgentHangsChild::TestUrgentHangsChild()
+{
+    MOZ_COUNT_CTOR(TestUrgentHangsChild);
+}
+
+TestUrgentHangsChild::~TestUrgentHangsChild()
+{
+    MOZ_COUNT_DTOR(TestUrgentHangsChild);
+}
+
+} // namespace _ipdltest
+} // namespace mozilla

ipc/ipdl/test/cxx/TestUrgentHangs.h

@@ -0,0 +1,66 @@
+#ifndef mozilla__ipdltest_TestUrgentHangs_h
+#define mozilla__ipdltest_TestUrgentHangs_h 1
+
+#include "mozilla/_ipdltest/IPDLUnitTests.h"
+
+#include "mozilla/_ipdltest/PTestUrgentHangsParent.h"
+#include "mozilla/_ipdltest/PTestUrgentHangsChild.h"
+
+namespace mozilla {
+namespace _ipdltest {
+
+
+class TestUrgentHangsParent :
+    public PTestUrgentHangsParent
+{
+public:
+    TestUrgentHangsParent();
+    virtual ~TestUrgentHangsParent();
+
+    static bool RunTestInProcesses() { return true; }
+    static bool RunTestInThreads() { return false; }
+
+    void Main();
+    void FinishTesting();
+
+    bool RecvTest1_2();
+    bool RecvTestInner();
+
+    bool ShouldContinueFromReplyTimeout() MOZ_OVERRIDE
+    {
+      return false;
+    }
+    virtual void ActorDestroy(ActorDestroyReason why) MOZ_OVERRIDE
+    {
+        passed("ok");
+        QuitParent();
+    }
+};
+
+
+class TestUrgentHangsChild :
+    public PTestUrgentHangsChild
+{
+public:
+    TestUrgentHangsChild();
+    virtual ~TestUrgentHangsChild();
+
+    bool RecvTest1_1();
+    bool RecvTest1_3();
+    bool RecvTest2();
+    bool RecvTest3();
+    bool RecvTest4();
+    bool RecvTest4_1();
+
+    virtual void ActorDestroy(ActorDestroyReason why) MOZ_OVERRIDE
+    {
+        QuitChild();
+    }
+};
+
+
+} // namespace _ipdltest
+} // namespace mozilla
+
+
+#endif // ifndef mozilla__ipdltest_TestUrgentHangs_h

ipc/ipdl/test/cxx/moz.build

@@ -46,6 +46,7 @@ SOURCES += [
     'TestSyncHang.cpp',
     'TestSyncWakeup.cpp',
     'TestUrgency.cpp',
+    'TestUrgentHangs.cpp',
 ]
 
 if CONFIG['OS_ARCH'] == 'Linux':
@@ -120,6 +121,7 @@ IPDL_SOURCES += [
     'PTestSyncWakeup.ipdl',
     'PTestSysVShmem.ipdl',
     'PTestUrgency.ipdl',
+    'PTestUrgentHangs.ipdl',
 ]
 
 include('/ipc/chromium/chromium-config.mozbuild')

testing/mochitest/mochitest_options.py

@@ -809,7 +809,7 @@ class B2GOptions(MochitestOptions):
         defaults["testPath"] = ""
         defaults["extensionsToExclude"] = ["specialpowers"]
         # See dependencies of bug 1038943.
-        defaults["defaultLeakThreshold"] = 5308
+        defaults["defaultLeakThreshold"] = 5404
         self.set_defaults(**defaults)
 
     def verifyRemoteOptions(self, options):