[Webkit-unassigned] [Bug 127029] [iOS] SubresourceLoader::didFinishLoading ASSERT(!m_resource->errorOccurred()); loading apple.com

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=127029





--- Comment #3 from Daniel Bates <dbates at webkit.org>  2014-01-21 10:46:42 PST ---
(In reply to comment #2)
> Would WebKit hit this assertion on Mac if there is a true decode error (e.g. garbage bytes instead of an image resource)? 

No.

> If not, what prevents that from happening?

Either decoding happens after the resource is loaded (e.g. a font) or, in the case of a CachedImage, we don't explicitly query the status of the image decoder and hence don't update the status of the CachedResource (i.e. CachedResource::setStatus()) when an actual decoder error occurs. CachedImage is implicitly (*) informed about a decode error by the return value of calling Image::setData() in CachedImage::addIncrementalDataBuffer(), <http://trac.webkit.org/browser/trunk/Source/WebCore/loader/cache/CachedImage.cpp?rev=161309#L406> and CachedImage::finishLoading(), <http://trac.webkit.org/browser/trunk/Source/WebCore/loader/cache/CachedImage.cpp?rev=161309#L447>. As aforementioned, we don't use this return value to update the status of the resource.

(*) Looking at the control flow for a bitmap image, Image::setData() => BitmapImage::dataChanged() => ImageSource::isSizeAvailable() => BitmapImage::isSizeAvailable() => ImageSource::isSizeAvailable(). And ImageSource::isSizeAvailable() (as defined in ImageSourceCG.cpp) will return false if the status of the decoder is EOF was encountered unexpectedly (kCGImageStatusUnexpectedEOF), invalid data (kCGImageStatusInvalidData), the image is an unknown type (kCGImageStatusUnknownType), the decoder is in the process of reading the header of the image (kCGImageStatusReadingHeader), and "the operation is not complete" (kCGImageStatusIncomplete). See <https://developer.apple.com/library/ios/documentation/graphicsimaging/reference/CGImageSource/Reference/reference.html> for more details on the aforementioned status codes.

> That said, it seems strange that a platform limitation like this is implemented via simulating a load failure.

This platform limitation doesn't trigger a load failure therefore the assertion failure. That is, when a decode error occurs we continue the load through to completion and hence SubresourceLoader::didFinishLoading() is called.

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