Title: [236487] trunk
Revision
236487
Author
[email protected]
Date
2018-09-25 16:29:34 -0700 (Tue, 25 Sep 2018)

Log Message

Unreviewed, rolling out r236420 and r236458.

This change is not fully correct

Reverted changesets:

"Do not do early processing of incoming sync IPC unless we're
waiting for a sync IPC reply"
https://bugs.webkit.org/show_bug.cgi?id=186941
https://trac.webkit.org/changeset/236420

"Unreviewed, update bug number of http/tests/misc/blob-
size.html crash on Windows"
https://trac.webkit.org/changeset/236458

Modified Paths

Removed Paths

Diff

Modified: trunk/LayoutTests/ChangeLog (236486 => 236487)


--- trunk/LayoutTests/ChangeLog	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/LayoutTests/ChangeLog	2018-09-25 23:29:34 UTC (rev 236487)
@@ -1,3 +1,20 @@
+2018-09-25  Chris Dumez  <[email protected]>
+
+        Unreviewed, rolling out r236420 and r236458.
+
+        This change is not fully correct
+
+        Reverted changesets:
+
+        "Do not do early processing of incoming sync IPC unless we're
+        waiting for a sync IPC reply"
+        https://bugs.webkit.org/show_bug.cgi?id=186941
+        https://trac.webkit.org/changeset/236420
+
+        "Unreviewed, update bug number of http/tests/misc/blob-
+        size.html crash on Windows"
+        https://trac.webkit.org/changeset/236458
+
 2018-09-25  Justin Fan  <[email protected]>
 
         WebGL 2 conformance: expected test results for the deqp suite (mac-mojave-wk2)

Deleted: trunk/LayoutTests/http/tests/misc/blob-size-expected.txt (236486 => 236487)


--- trunk/LayoutTests/http/tests/misc/blob-size-expected.txt	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/LayoutTests/http/tests/misc/blob-size-expected.txt	2018-09-25 23:29:34 UTC (rev 236487)
@@ -1,10 +0,0 @@
-Tests that constructing a Blob and querying its size right away is not flaky
-
-On success, you will see a series of "PASS" messages, followed by "TEST COMPLETE".
-
-
-PASS Getting the Blob size is not flaky
-PASS successfullyParsed is true
-
-TEST COMPLETE
-

Deleted: trunk/LayoutTests/http/tests/misc/blob-size.html (236486 => 236487)


--- trunk/LayoutTests/http/tests/misc/blob-size.html	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/LayoutTests/http/tests/misc/blob-size.html	2018-09-25 23:29:34 UTC (rev 236487)
@@ -1,26 +0,0 @@
-<!DOCTYPE html>
-<html>
-<body>
-<script src=""
-<script>
-description("Tests that constructing a Blob and querying its size right away is not flaky");
-
-function runTest() {
-    for (let i = 0; i < 50; i++) {
-        for (let j = 0; j < 200; j++) {
-            fetch('/resources/square100.png', { cache: 'no-store' }).then(function(res) { });
-        }
-        for (let k = 0; k < 5; k++) {
-        if ((new Blob(["aaaa"])).size != 4) {
-            testFailed("Getting the Blob size is flaky");
-            return;
-        }
-        }
-    }
-    testPassed("Getting the Blob size is not flaky");
-}
-
-runTest();
-</script>
-</body>
-</html>

Modified: trunk/LayoutTests/platform/win/TestExpectations (236486 => 236487)


--- trunk/LayoutTests/platform/win/TestExpectations	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/LayoutTests/platform/win/TestExpectations	2018-09-25 23:29:34 UTC (rev 236487)
@@ -4173,8 +4173,6 @@
 
 webkit.org/b/188948 fast/events/dblclick-event-getModifierState.html [ Failure ]
 
-webkit.org/b/189948 http/tests/misc/blob-size.html [ Crash ]
-
 # Tests failing on EWS after updating to iTunes 12.8
 webkit.org/b/189194 http/tests/IndexedDB/collect-IDB-objects.https.html [ Failure ]
 webkit.org/b/189194 http/tests/referrer-policy/origin-when-cross-origin/cross-origin-http.https.html [ Failure ]

Modified: trunk/Source/WebKit/ChangeLog (236486 => 236487)


--- trunk/Source/WebKit/ChangeLog	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/Source/WebKit/ChangeLog	2018-09-25 23:29:34 UTC (rev 236487)
@@ -1,3 +1,20 @@
+2018-09-25  Chris Dumez  <[email protected]>
+
+        Unreviewed, rolling out r236420 and r236458.
+
+        This change is not fully correct
+
+        Reverted changesets:
+
+        "Do not do early processing of incoming sync IPC unless we're
+        waiting for a sync IPC reply"
+        https://bugs.webkit.org/show_bug.cgi?id=186941
+        https://trac.webkit.org/changeset/236420
+
+        "Unreviewed, update bug number of http/tests/misc/blob-
+        size.html crash on Windows"
+        https://trac.webkit.org/changeset/236458
+
 2018-09-25  Jiewen Tan  <[email protected]>
 
         Unreviewed, a quick fix after r236481.

Modified: trunk/Source/WebKit/Platform/IPC/Connection.cpp (236486 => 236487)


--- trunk/Source/WebKit/Platform/IPC/Connection.cpp	2018-09-25 23:21:25 UTC (rev 236486)
+++ trunk/Source/WebKit/Platform/IPC/Connection.cpp	2018-09-25 23:29:34 UTC (rev 236487)
@@ -719,13 +719,8 @@
     // Check if this is a sync message or if it's a message that should be dispatched even when waiting for
     // a sync reply. If it is, and we're waiting for a sync reply this message needs to be dispatched.
     // If we don't we'll end up with a deadlock where both sync message senders are stuck waiting for a reply.
-    {
-        LockHolder locker(m_syncReplyStateMutex);
-        // If the m_onlySendMessagesAsDispatchWhenWaitingForSyncReplyWhenProcessingSuchAMessage flag is set, then our sync messages will not be processed by the destination process while it is waiting for a sync IPC reply.
-        // As a result, we need to process incoming sync messages early here, even if we're not waiting for a sync IPC reply in order to prevent deadlocks.
-        if ((m_onlySendMessagesAsDispatchWhenWaitingForSyncReplyWhenProcessingSuchAMessage || !m_pendingSyncReplies.isEmpty()) && SyncMessageState::singleton().processIncomingMessage(*this, message))
-            return;
-    }
+    if (SyncMessageState::singleton().processIncomingMessage(*this, message))
+        return;
 
     // Check if we're waiting for this message.
     {
_______________________________________________
webkit-changes mailing list
[email protected]
https://lists.webkit.org/mailman/listinfo/webkit-changes

Reply via email to