[webkit-dev] Fwd: Understending LLInt

[wingoog moon]

---------- Forwarded message ----------
From: wingoog moon <wingoog91 at gmail.com>
Date: Wed, Nov 14, 2012 at 1:51 AM
Subject: Understending LLInt
To: squirrelfish-dev at lists.webkit.org


Hi All.

At which point LLInt starts to compile bytecode to the native code?
If I'm not mistaken it should be in prepareForExecution function, so I
guess that this part of code actually does translation from bytecode to
native code.

 if (JITCode::isBaselineCode(jitType)) {
        // Start off in the low level interpreter.
        LLInt::getEntrypoint(exec->globalData(), codeBlock.get(), jitCode);
        codeBlock->setJITCode(jitCode, MacroAssemblerCodePtr());
        return true;
    }

But I just cannot figure out how translation actually is done. My only
assumption that this function only set up some native code, and actually
compilation begins when calling functions from native assembly code.

Can anyone please explain LLInt working mechanism, any links or articles?
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.webkit.org/pipermail/webkit-dev/attachments/20121114/18df8f15/attachment.html>