[Webkit-unassigned] [Bug 146448] JavaScriptCore performance is very bad on Windows

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=146448

Mark Lam <mark.lam at apple.com> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #256896|review?                     |review-
              Flags|                            |

--- Comment #26 from Mark Lam <mark.lam at apple.com> ---
Comment on attachment 256896
  --> https://bugs.webkit.org/attachment.cgi?id=256896
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=256896&action=review

Per, there are 2 changes here:
1. Replacing use of std::thread::id with WTF::ThreadIdentifier.
2. Optimizing out the computation of stack bounds when still in the same Windows fiber.

Can you (or someone) please confirm that (1) actually contributes to the performance difference you’re seeing?  These 2 changes should be tested separately to measure their effects.

> Source/JavaScriptCore/runtime/JSLock.cpp:121
> +        m_ownerThreadID = WTF::currentThread();

No need to use the WTF::.  The convention for WTF APIs is that their header files already provide "using” declarations.  We can just use an unqualified currentThread() here.

> Source/JavaScriptCore/runtime/JSLock.cpp:200
> +    ASSERT(!m_hasExclusiveThread || (exclusiveThread() == WTF::currentThread()));

Ditto.  No need for WTF:: qualifier.

> Source/JavaScriptCore/runtime/JSLock.cpp:203
> +    return m_ownerThreadID == WTF::currentThread();

Ditto.

> Source/JavaScriptCore/runtime/JSLock.cpp:210
> +        ASSERT(exclusiveThread() == WTF::currentThread());

Ditto.

> Source/WebCore/bindings/js/JSDOMWindowBase.cpp:206
> +        vm->setExclusiveThread(WTF::currentThread());

Ditto.  No need to WTF:: qualifier.

-- 
You are receiving this mail because:
You are the assignee for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.webkit.org/pipermail/webkit-unassigned/attachments/20150716/e456b163/attachment.html>