Bug 149706 – Clean up PassRefPtr in websockets

• Status: RESOLVED
• Resolution: WONTFIX
• Component: New Bugs

[Gyuyoung Kim]

SSIA

[Gyuyoung Kim]

Created attachment 262251 [details]
Patch

[Brent Fulgham]

What happened to this? Seems like it was a useful in-progress change.

[Gyuyoung Kim]

(In reply to comment #2)
> What happened to this? Seems like it was a useful in-progress change.

Oops, I have forgotten this bug. Let me try to upload new patch based on trunk.

[Gyuyoung Kim]

Created attachment 280069 [details]
Patch

[Build Bot]

Comment on attachment 280069 [details]
Patch

Attachment 280069 [details] did not pass mac-wk2-ews (mac-wk2):
Output: http://webkit-queues.webkit.org/results/1405461

Number of test failures exceeded the failure limit.

[Build Bot]

Created attachment 280071 [details]
Archive of layout-test-results from ews106 for mac-yosemite-wk2

The attached test failures were seen while running run-webkit-tests on the mac-wk2-ews.
Bot: ews106  Port: mac-yosemite-wk2  Platform: Mac OS X 10.10.5

[Gyuyoung Kim]

Created attachment 280116 [details]
Patch

[Chris Dumez]

Comment on attachment 280116 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=280116&action=review

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:56
> +    , m_bridge(Bridge::create(m_workerClientWrapper.copyRef(), m_workerGlobalScope.get(), taskMode))

Clearly, m_workerClientWrapper cannot be null here, given that it is initialized on the previous line.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:145
> +WorkerThreadableWebSocketChannel::Peer::Peer(ThreadableWebSocketChannelClientWrapper* clientWrapper, WorkerLoaderProxy& loaderProxy, ScriptExecutionContext* context, const String& taskMode)

I don't think clientWrapper can be null so it looks like this should be a reference. Given that We ref it, I think this could even be a Ref<>&&.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:350
> +WorkerThreadableWebSocketChannel::Bridge::Bridge(RefPtr<ThreadableWebSocketChannelClientWrapper>&& workerClientWrapper, WorkerGlobalScope* workerGlobalScope, const String& taskMode)

Looks like this should be a Ref<>&& for the workerClientWrapper, given the assertion below.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:365
> +void WorkerThreadableWebSocketChannel::Bridge::mainThreadInitialize(ScriptExecutionContext& context, WorkerLoaderProxy* loaderProxy, ThreadableWebSocketChannelClientWrapper* clientWrapper, const String& taskMode)

I don't think clientWrapper can be null so it should probably be a reference.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:-371
> -    RefPtr<ThreadableWebSocketChannelClientWrapper> clientWrapper = prpClientWrapper;

It is important to ref it for the capture in the lambda below. The new code looks unsafe.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.cpp:403
> +        mainThreadInitialize(context, loaderProxy, workerClientWrapper.get(), capturedTaskMode.string());

I don't think workerClientWrapper can be null here given that is it dereferenced below.

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.h:82
> +        static Peer* create(ThreadableWebSocketChannelClientWrapper* clientWrapper, WorkerLoaderProxy& loaderProxy, ScriptExecutionContext* context, const String& taskMode)

Why does this return a raw pointer? This should really return a Ref<> (although let's do this separately).

> Source/WebCore/Modules/websockets/WorkerThreadableWebSocketChannel.h:128
> +        static Ref<Bridge> create(RefPtr<ThreadableWebSocketChannelClientWrapper>&& workerClientWrapper, WorkerGlobalScope* workerGlobalScope, const String& taskMode)

workerClientWrapper cannot be null, to I would prefer a Ref<>&&.

[Build Bot]

Comment on attachment 280116 [details]
Patch

Attachment 280116 [details] did not pass mac-ews (mac):
Output: http://webkit-queues.webkit.org/results/1410046

Number of test failures exceeded the failure limit.

[Build Bot]

Created attachment 280117 [details]
Archive of layout-test-results from ews100 for mac-yosemite

The attached test failures were seen while running run-webkit-tests on the mac-ews.
Bot: ews100  Port: mac-yosemite  Platform: Mac OS X 10.10.5

[Gyuyoung Kim]

Created attachment 280118 [details]
Patch

[Ahmad Saleem]

PassRefPtr usage does not exist anymore in Webkit GitHub as of today. So I am marking this as "RESOLVED WONTFIX". Thanks!