[Webkit-unassigned] [Bug 58123] New: Document source preload scanned repeatedly

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=58123

           Summary: Document source preload scanned repeatedly
           Product: WebKit
           Version: 528+ (Nightly build)
          Platform: PC
        OS/Version: Mac OS X 10.5
            Status: NEW
          Severity: Normal
          Priority: P2
         Component: Page Loading
        AssignedTo: webkit-unassigned at lists.webkit.org
        ReportedBy: koivisto at iki.fi


Clearing the preload scanner after parsing resumes (from http://trac.webkit.org/changeset/63154)

void HTMLDocumentParser::resumeParsingAfterScriptExecution()
{
    ASSERT(!inScriptExecution());
    ASSERT(!m_treeBuilder->isPaused());

    m_preloadScanner.clear()
...

makes us rescan the document text multiple times if the parser gets blocked repeatedly during loading. For example when reloading nytimes.com, we scan almost the entire 120kb main document 5 times, causing us to spent way more time in the preload scanner than the main tokenizer.

This happens because when the scanner is cleared we lose information about how far we have already scanned. When the parser gets blocked again, the preload scanner is re-constructed and the scanning starts from the current position of the main parser:

void HTMLDocumentParser::pumpTokenizer(SynchronousMode mode)
{
...
    if (isWaitingForScripts()) {
        ASSERT(m_tokenizer->state() == HTMLTokenizer::DataState);
        if (!m_preloadScanner) {
            m_preloadScanner.set(new HTMLPreloadScanner(document()));
            m_preloadScanner->appendToEnd(m_input.current());
        }
        m_preloadScanner->scan();
...

If the blocking happens early in the document (as it often does) most of the document text gets re-scanned.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.