Bug 76969

Summary:

Make DFG update topCallFrame

Product:

WebKit

Reporter:

Oliver Hunt <oliver>

Component:

New Bugs

Assignee:

Oliver Hunt <oliver>

Status:

RESOLVED FIXED

Severity:

Normal

Priority:

Version:

528+ (Nightly build)

Hardware:

Unspecified

OS:

Unspecified

Attachments:

Description	Flags
Patch	none
Patch	fpizlo: review+

Description Oliver Hunt 2012-01-24 17:53:10 PST

Make DFG update topCallFrame

Comment 1 Oliver Hunt 2012-01-24 17:56:09 PST

Created attachment 123861 [details]
Patch

Comment 2 Oliver Hunt 2012-01-24 17:57:48 PST

Created attachment 123862 [details]
Patch

Comment 3 Filip Pizlo 2012-01-25 11:40:32 PST

Comment on attachment 123862 [details]
Patch

r=me.

Comment 4 Oliver Hunt 2012-01-25 11:43:41 PST

Committed r105905: <http://trac.webkit.org/changeset/105905>

Comment 5 Geoffrey Garen 2012-01-25 11:44:43 PST

Comment on attachment 123862 [details]
Patch

View in context: https://bugs.webkit.org/attachment.cgi?id=123862&action=review

Can you post performance results?

> Source/JavaScriptCore/interpreter/Interpreter.h:91
> +    class NativeCallFrameTracer {
> +    public:
> +        ALWAYS_INLINE NativeCallFrameTracer(JSGlobalData* global, CallFrame* callFrame)
> +        {
> +            ASSERT(global);
> +            ASSERT(callFrame);
> +            global->topCallFrame = callFrame;
> +        }
> +    };

Is it OK that we don't reset topCallFrame on destruction? When do we need to reset it, and when can we skip that step?

I don't think the words "native" and "tracer" are clear enough to distinguish NativeCallFrameTracer from TopCallFrameSetter. Answers to the above questions can provide us with a better name.