Bug 13323 – REGRESSION: 1 leak seen after r20813

• Status: NEW
• Component: Page Loading

[Geoffrey Garen]

The error text (http://build.webkit.org/post-commit-leaks-powerpc-mac-os-x/builds/5547/step-leak-test/0): 
ERROR: No parent frame in transitionToCommitted:, FrameLoadTypeInternal
(/Users/buildbot/Desktop/BuildData/WebKit-BuildSlave/post-commit-leaks-powerpc-mac-os-x/build/WebCore/loader/FrameLoader.cpp:4053 updateHistoryForInternalLoad)

The leak (http://build.webkit.org/results/post-commit-leaks-powerpc-mac-os-x/5547/DumpRenderTree-leaks.txt):

Process 6214: 47151 nodes malloced for 8824 KB
Process 6214: 8 leaks for 1936 total leaked bytes.
Leak: 0x028ec400  size=1024	instance of 'NSCFString'	
	0xa07b9674 0x00010788 0x0000021e 0x66696c65 	.{.t........file
	0x3a2f2f2f 0x55736572 0x732f6275 0x696c6462 	:///Users/buildb
	0x6f742f44 0x65736b74 0x6f702f42 0x75696c64 	ot/Desktop/Build
	0x44617461 0x2f576562 0x4b69742d 0x4275696c 	Data/WebKit-Buil
	0x64536c61 0x76652f70 0x6f73742d 0x636f6d6d 	dSlave/post-comm
	0x69742d6c 0x65616b73 0x2d706f77 0x65727063 	it-leaks-powerpc
	0x2d6d6163 0x2d6f732d 0x782f6275 0x696c642f 	-mac-os-x/build/
	0x4c61796f 0x75745465 0x7374732f 0x66617374 	LayoutTests/fast
	...
	Call stack: [thread 13]: | 0x3 | start | _start | main | dumpRenderTree | runTest | -[NSRunLoop runMode:beforeDate:] | CFRunLoopRunSpecific | __CFRunLoopRun | __CFRunLoopDoSources0 | _sendCallbacks | -[NSURLConnection(NSURLConnectionInternal) _sendCallbacks] | -[NSURLConnection(NSURLConnectionInternal) _sendDidReceiveDataCallback] | -[WebCoreResourceHandleAsDelegate connection:didReceiveData:lengthReceived:] | WebCore::ResourceLoader::didReceiveData(WebCore::ResourceHandle*, char const*, int, int) | WebCore::MainResourceLoader::didReceiveData(char const*, int, long long, bool) | WebCore::ResourceLoader::didReceiveData(char const*, int, long long, bool) | WebCore::MainResourceLoader::addData(char const*, int, bool) | WebCore::FrameLoader::receivedData(char const*, int) | WebCore::DocumentLoader::receivedData(char const*, int) | WebCore::DocumentLoader::commitLoad(char const*, int) | WebCore::DocumentLoader::commitIfReady() | WebCore::FrameLoader::commitProvisionalLoad(WTF::PassRefPtr<WebCore::PageCache>) | WebCore::FrameLoader::transitionToCommitted(WTF::PassRefPtr<WebCore::PageCache>) | WebCore::FrameLoader::updateHistoryForStandardLoad() | WebCore::FrameLoader::updateGlobalHistoryForStandardLoad(WebCore::KURL const&) | WebFrameLoaderClient::updateGlobalHistoryForStandardLoad(WebCore::KURL const&) | WebCore::KURL::getNSURL() const | WebCore::KURL::createCFURL() const | CFURLCreateAbsoluteURLWithBytes | CFStringCreateWithBytes | __CFStringCreateImmutableFunnel3 | _CFRuntimeCreateInstance | CFAllocatorAllocate 
7 leaks excluded (not printed)

[Brady Eidson]

The leak is not a result of 20813 - it first appeared in build 5547 and appears to be tied to 20815.

The error messages... well, I missed those in my final run of layout tests before committing this patch.  
I will explore them soon-ish.

[Darin Adler]

So these should be two separate bugs.

[Brady Eidson]

This bug is now just for the leak.  I have filed http://bugs.webkit.org/show_bug.cgi?id=13330 for the error logging

[Brady Eidson]

That said, I could not reproduce the leak locally, and it seems only one of the bots can reproduce it - and intermittently at that!  I don't know how to explore this - and its difficult to say for certain it was 20813 since the leak could have been there before but that bot just wasn't running.

[Darin Adler]

<rdar://problem/5153016>

[Oliver Hunt]

A single intermittent leak does not a p1 make