Clear pointers to PendingScript when dispose()d in pendingScriptFinished()

third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp

Index: third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp
diff --git a/third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp b/third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp
index 88afa39d916a2a36115136ef176f2b1000c90df6..596bb2d50b30d3c0c0e8d8874bc392daa6ac5dec 100644
--- a/third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp
+++ b/third_party/WebKit/Source/core/html/parser/HTMLParserScriptRunner.cpp
@@ -363,6 +363,13 @@ void HTMLParserScriptRunner::pendingScriptFinished(
   // The parser is unprepared to be told, and doesn't need to be.
   if (isExecutingScript() && pendingScript->resource()->wasCanceled()) {
     pendingScript->dispose();
+
+    // TODO(hiroshige): According to the comment above,
+    // this block is intended to process |m_parserBlockingScript|.
+    // If this CHECK() should fail, implement the case where
+    // |m_parserBlockingScript == m_scriptsToExecuteAfterParsing.first()|.

[kouhei (in TOK), 2017/02/28 00:41:48]

I'm not sure if we should do that per se. I'd rather rm member
m_parserBlockingScript and unify to one queue (possibly
m_scriptsToExecuteAfterParsing)

[hiroshige, 2017/02/28 00:48:55]

|m_parserBlockingScript| and |m_scriptsToExecuteAfterParsing| has separate
concepts in the spec, as commented in HTMLParserScriptRunner.h, so having these
two separately makes the code directly corresponds to the spec.

(It might be nice if we can unify these two also in the spec. I feel the
assumptions/invariants about these two are not so clear in impl or spec)

+    CHECK_EQ(pendingScript, parserBlockingScript());
+    m_parserBlockingScript = nullptr;
     return;
   }