Bug 179699 – Fix a bit-rotted Interpreter::dumpRegisters() and make it more robust.

• Status: RESOLVED
• Resolution: FIXED
• Component: JavaScriptCore

[Mark Lam]

<rdar://problem/35462346>

[Mark Lam]

Created attachment 326935 [details]
proposed patch.

[Mark Lam]

Created attachment 326936 [details]
proposed patch.

[Michael Saboff]

Comment on attachment 326936 [details]
proposed patch.

View in context: https://bugs.webkit.org/attachment.cgi?id=326936&action=review

r=me after reversing the order of CallerFrame and ReturnPC.

> Source/JavaScriptCore/interpreter/Interpreter.cpp:-441
> +    dataLogF("[CallerFrame]              | %10p | %p \n", it, callFrame->callerFrame());
>      --it;
>  #if ENABLE(JIT)
>      AbstractPC pc = callFrame->abstractReturnPC(callFrame->vm());
>      if (pc.hasJITReturnAddress())
>          dataLogF("[ReturnJITPC]              | %10p | %p \n", it, pc.jitReturnAddress().value());
> -#endif

I think you need to flip the order of CallerFrame and ReturnJITPC as I'm pretty sure that the CallerFrame (aka previous frame pointer) is pushed on after the Return PC (or at a lower address for ARM64 as they are pushed simultaneously).

Also, change ReturnJITPC to ReturnPC as it may not be a JIT'ed PC.

> Source/JavaScriptCore/interpreter/Interpreter.cpp:465
> +                ? "INVALID"

Nit - I'd make this "Unknown" or leave it blank.

[Mark Lam]

Created attachment 326942 [details]
patch for landing.

Thanks for the review.

[Mark Lam]

Landed in r224862: <http://trac.webkit.org/r224862>.

[Mark Lam]

Landed a CLoop build fix (after r224862) in r224868: <http://trac.webkit.org/r224868>.