Bug 48561

Summary: Assertion failure in DocumentLoader::commitData when loading a media document in WebKit2
Product: WebKit Reporter: Adam Roben (:aroben) <aroben>
Component: Page LoadingAssignee: Nobody <webkit-unassigned>
Status: RESOLVED FIXED    
Severity: Normal CC: andersca, sam, webkit.review.bot
Priority: P2 Keywords: InRadar
Version: 528+ (Nightly build)   
Hardware: All   
OS: All   
Attachments:
Description Flags
Cancel main resource loads after we hand them off to the media engine eric.carlson: review+

Adam Roben (:aroben)
Reported 2010-10-28 13:54:24 PDT
This is the WebKit2 equivalent of bug 48531. To reproduce: 1. Go to http://www.nch.com.au/acm/sample.wav in a debug build You'll hit an assertion in DocumentLoader::commitData: ASSERT(m_frame->document()->parsing()); The wav file seems to load fine in Release builds. Here's the backtrace: > WebKit.dll!WebCore::DocumentLoader::commitData(const char * bytes=0x0524f010, int length=67868) Line 306 + 0x35 bytes C++ WebKit.dll!WebKit::WebFrameLoaderClient::committedLoad(WebCore::DocumentLoader * loader=0x020d95b8, const char * data=0x0524f010, int length=67868) Line 681 C++ WebKit.dll!WebCore::DocumentLoader::commitLoad(const char * data=0x0524f010, int length=67868) Line 292 + 0x29 bytes C++ WebKit.dll!WebCore::DocumentLoader::receivedData(const char * data=0x0524f010, int length=67868) Line 320 C++ WebKit.dll!WebCore::MainResourceLoader::addData(const char * data=0x0524f010, int length=67868, bool allAtOnce=false) Line 157 C++ WebKit.dll!WebCore::ResourceLoader::didReceiveData(const char * data=0x0524f010, int length=67868, __int64 lengthReceived=67868, bool allAtOnce=false) Line 262 + 0x1b bytes C++ WebKit.dll!WebCore::MainResourceLoader::didReceiveData(const char * data=0x0524f010, int length=67868, __int64 lengthReceived=67868, bool allAtOnce=false) Line 437 C++ WebKit.dll!WebCore::ResourceLoader::didReceiveData(WebCore::ResourceHandle * __formal=0x02052460, const char * data=0x0524f010, int length=67868, int lengthReceived=67868) Line 415 + 0x1f bytes C++ WebKit.dll!WebCore::didReceiveData(_CFURLConnection * conn=0x020e2d88, const __CFData * data=0x0523d348, long originalLength=67868, const void * clientInfo=0x02052460) Line 214 + 0x2a bytes C++ CFNetwork.dll!URLConnectionClient::_clientDidReceiveData() Note that on Mac you'll instead hit a crash (bug 48559), but if that crash didn't happen this would.
Attachments
Cancel main resource loads after we hand them off to the media engine (7.36 KB, patch)
2010-11-01 09:24 PDT, Adam Roben (:aroben)
eric.carlson: review+
Adam Roben (:aroben)
Comment 1 2010-10-28 13:54:53 PDT
Adam Roben (:aroben)
Comment 2 2010-11-01 09:24:49 PDT
Created attachment 72516 [details] Cancel main resource loads after we hand them off to the media engine
WebKit Review Bot
Comment 3 2010-11-01 09:27:18 PDT
Attachment 72516 [details] did not pass style-queue: Failed to run "['WebKitTools/Scripts/check-webkit-style']" exit_code: 1 WebKit2/WebProcess/WebCoreSupport/mac/WebErrorsMac.mm:174: Line contains tab character. [whitespace/tab] [5] Total errors found: 1 in 6 files If any of these errors are false positives, please file a bug against check-webkit-style.
Eric Carlson
Comment 4 2010-11-01 09:40:25 PDT
Comment on attachment 72516 [details] Cancel main resource loads after we hand them off to the media engine View in context: https://bugs.webkit.org/attachment.cgi?id=72516&action=review > WebKit2/WebProcess/WebCoreSupport/qt/WebErrorsQt.cpp:93 > +ResourceError pluginWillHandleLoadError(const ResourceResponse&) > +{ > + notImplemented(); > + return ResourceError(); > +} Might be worth filing a bug about this.
Adam Roben (:aroben)
Comment 5 2010-11-01 09:44:42 PDT
Comment on attachment 72516 [details] Cancel main resource loads after we hand them off to the media engine View in context: https://bugs.webkit.org/attachment.cgi?id=72516&action=review >> WebKit2/WebProcess/WebCoreSupport/qt/WebErrorsQt.cpp:93 >> +} > > Might be worth filing a bug about this. Filed bug 48764.
Adam Roben (:aroben)
Comment 6 2010-11-01 09:52:24 PDT
Note You need to log in before you can comment on or make changes to this bug.