[webkit-reviews] review granted: [Bug 110075] REGRESSION: Crash in MainResourceLoader::setDataBufferingPolicy when sharing an html5 video via email : [Attachment 189426] Fix ChangeLog

[bugzilla-daemon at webkit.org]

Alexey Proskuryakov <ap at webkit.org> has granted Nate Chapin
<japhet at chromium.org>'s request for review:
Bug 110075: REGRESSION: Crash in MainResourceLoader::setDataBufferingPolicy
when sharing an html5 video via email
https://bugs.webkit.org/show_bug.cgi?id=110075

Attachment 189426: Fix ChangeLog
https://bugs.webkit.org/attachment.cgi?id=189426&action=review

------- Additional Comments from Alexey Proskuryakov <ap at webkit.org>
> The iframe in the webarchive points to a 0-byte mpeg. I'd guess that's the
cause, but I'm not 100% sure.

This looks more like a bug than like correct behavior. This happens because we
cancel main resource load when switching to a media document, and that somehow
translates into a frame loader failure. But it's a resource load cancellation,
not a frame load failure.


    // If the document is a stand-alone media document, now is the right time
to cancel the WebKit load.
    // FIXME: This code should be shared across all ports.
<http://webkit.org/b/48762>.
    if (m_frame->coreFrame()->document()->isMediaDocument())
       
loader->cancelMainResourceLoad(pluginWillHandleLoadError(loader->response()));


#0	0x000000010128afcc in
WebKit::WebFrameLoaderClient::dispatchDidFailLoading(WebCore::DocumentLoader*,
unsigned long, WebCore::ResourceError const&) at
/Users/ap/Safari/OpenSource/Source/WebKit2/WebProcess/WebCoreSupport/WebFrameLo
aderClient.cpp:249
#1	0x00000001046df8af in
WebCore::MainResourceLoader::receivedError(WebCore::ResourceError const&) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:99
#2	0x00000001046dfb27 in
WebCore::MainResourceLoader::cancel(WebCore::ResourceError const&) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:132
#3	0x000000010399a11f in
WebCore::DocumentLoader::cancelMainResourceLoad(WebCore::ResourceError const&)
at /Users/ap/Safari/OpenSource/Source/WebCore/loader/DocumentLoader.cpp:913
#4	0x000000010128db82 in
WebKit::WebFrameLoaderClient::committedLoad(WebCore::DocumentLoader*, char
const*, int) at
/Users/ap/Safari/OpenSource/Source/WebKit2/WebProcess/WebCoreSupport/WebFrameLo
aderClient.cpp:852
#5	0x0000000103997b20 in WebCore::DocumentLoader::commitLoad(char const*,
int) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/DocumentLoader.cpp:318
#6	0x00000001039980cb in WebCore::DocumentLoader::receivedData(char
const*, int) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/DocumentLoader.cpp:397
#7	0x00000001046e1f4a in
WebCore::MainResourceLoader::dataReceived(WebCore::CachedResource*, char
const*, int) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:522
#8	0x00000001046e0e20 in
WebCore::MainResourceLoader::continueAfterContentPolicy(WebCore::PolicyAction,
WebCore::ResourceResponse const&) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:365
#9	0x00000001046e1266 in
WebCore::MainResourceLoader::continueAfterContentPolicy(WebCore::PolicyAction)
at /Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:381

#10	0x00000001046e11ab in
WebCore::MainResourceLoader::callContinueAfterContentPolicy(void*,
WebCore::PolicyAction) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:373
#11	0x00000001046e18c6 in
WebCore::MainResourceLoader::responseReceived(WebCore::CachedResource*,
WebCore::ResourceResponse const&) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:450
#12	0x00000001046df594 in
WebCore::MainResourceLoader::handleSubstituteDataLoadNow(WebCore::RunLoopTimer<
WebCore::MainResourceLoader>*) at
/Users/ap/Safari/OpenSource/Source/WebCore/loader/MainResourceLoader.cpp:628

I don't know if that's a long-standing behavior, or something that got broken
with MainResourceLoader refactoring recently. Would you be willing to take a
look?

r+ for this patch.