Changeset 93871 in webkit

Timestamp:

Aug 26, 2011, 6:15:00 AM (14 years ago)

Author:

gavinp@chromium.org

Message:

add a state enumeration to track down cause of null CachedScript execution
​https://bugs.webkit.org/show_bug.cgi?id=66939

Reviewed by Alexey Proskuryakov.

Over in ​http://code.google.com/p/chromium/issues/detail?id=75604 I
have a bug I cannot reproduce. I added an earlier crash in
​https://bugs.webkit.org/show_bug.cgi?id=65563 , and we've since
gotten many good stacks, all coming in through a failed request that
eventually calls notifyFinished() on a ScriptElement with a NULL
m_cachedScript.

I'd like to know how this got NULL. This enumeration should let
me find that in stack dumps from reproductions.

No new tests, as my goal here is to get telemetry on a bug that I cannot reproduce.

• dom/ScriptElement.cpp:

(WebCore::ScriptElement::ScriptElement):
(WebCore::ScriptElement::requestScript):
(WebCore::ScriptElement::stopLoadRequest):
(WebCore::ScriptElement::notifyFinished):

• dom/ScriptElement.h:

Location:

trunk/Source/WebCore

Files:

• ChangeLog (modified) (1 diff)
• dom/ScriptElement.cpp (modified) (4 diffs)
• dom/ScriptElement.h (modified) (1 diff)

trunk/Source/WebCore/ChangeLog

@@ +1 @@
+2011-08-26  Gavin Peters  <gavinp@chromium.org>
+
+        add a state enumeration to track down cause of null CachedScript execution
+        https://bugs.webkit.org/show_bug.cgi?id=66939
+
+        Reviewed by Alexey Proskuryakov.
+
+        Over in http://code.google.com/p/chromium/issues/detail?id=75604 I
+        have a bug I cannot reproduce.  I added an earlier crash in
+        https://bugs.webkit.org/show_bug.cgi?id=65563 , and we've since
+        gotten many good stacks, all coming in through a failed request that
+        eventually calls notifyFinished() on a ScriptElement with a NULL
+        m_cachedScript.
+
+        I'd like to know how this got NULL.  This enumeration should let
+        me find that in stack dumps from reproductions.
+
+        No new tests, as my goal here is to get telemetry on a bug that I cannot reproduce.
+
+        * dom/ScriptElement.cpp:
+        (WebCore::ScriptElement::ScriptElement):
+        (WebCore::ScriptElement::requestScript):
+        (WebCore::ScriptElement::stopLoadRequest):
+        (WebCore::ScriptElement::notifyFinished):
+        * dom/ScriptElement.h:
+
 2011-08-26  Mike Reed  <reed@google.com>
 

trunk/Source/WebCore/dom/ScriptElement.cpp

@@ -65 +65 @@
     , m_forceAsync(!parserInserted)
     , m_willExecuteInOrder(false)
+    , m_cachedScriptState(NeverSet)
 {
     ASSERT(m_element);
@@ -259 +260 @@
     m_isExternalScript = true;
 
-    if (m_cachedScript)
+    if (m_cachedScript) {
+        ASSERT(m_cachedScriptState == NeverSet);
+        m_cachedScriptState = Set;
         return true;
+    }
 
     dispatchErrorEvent();
@@ -296 +300 @@
         if (!m_willBeParserExecuted)
             m_cachedScript->removeClient(this);
+        ASSERT(m_cachedScriptState == Set);
+        m_cachedScriptState = ZeroedInStopLoadRequest;
         m_cachedScript = 0;
     }
@@ -321 +327 @@
     else
         m_element->document()->scriptRunner()->queueScriptForExecution(this, m_cachedScript, ScriptRunner::ASYNC_EXECUTION);
+
+    ASSERT(m_cachedScriptState == Set);
+    m_cachedScriptState = ZeroedInNotifyFinished;
     m_cachedScript = 0;
 }

trunk/Source/WebCore/dom/ScriptElement.h

@@ -105 +105 @@
     String m_characterEncoding;
     String m_fallbackCharacterEncoding;
+    
+    // Temporary: intended to help debug how we get notifyFinished() called when m_cachedScript is null,
+    // which seems to sometimes happen (see http://code.google.com/p/chromium/issues/detail?id=75604 )
+    enum {
+      NeverSet,
+      Set,
+      ZeroedInStopLoadRequest,
+      ZeroedInNotifyFinished,
+    } m_cachedScriptState;
 };