[Webkit-unassigned] [Bug 158083] LLInt should support other types of prototype GetById caching.

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sun Jul 31 17:34:15 PDT 2016


https://bugs.webkit.org/show_bug.cgi?id=158083

--- Comment #109 from Caio Lima <ticaiolima at gmail.com> ---
(In reply to comment #108)
> Comment on attachment 283306 [details]
> Patch
> 
> View in context:
> https://bugs.webkit.org/attachment.cgi?id=283306&action=review
> 
> > Source/JavaScriptCore/llint/LowLevelInterpreter32_64.asm:1478
> > +        loadEncodedThisValue(t3, a1)
> > +        push t5, cfr
> > +        loadp 20[PC], t5
> > +        # Inlining the GetValueFunc call
> > +        move t2, a3 # Load arg2 PropertyName
> > +        if BIG_ENDIAN
> > +            move a1, a2 # Load arg1 Payload of EncodedJSValue
> > +            move CellTag, a1 # Load arg1 Tag of EncodedJSValue
> > +        else
> > +            move CellTag, a2 # Load arg1 Tag of EncodedJSValue
> > +        end
> > +        move cfr, a0 # Loading exec
> > +        cCall4(t5)
> 
> Looks good. Although I still think you need to handle the EABI. For the EABI
> you'll need to push the PropertyName onto the stack put the EncodedJSValue
> into r2-r3. r1 should be empty.

Thank you to inform me about it. Which architectures use EABI? I suppose ARM, but I would like to be sure before send a new patch. Also, do you mind point me where you found this information?

> > Source/JavaScriptCore/llint/LowLevelInterpreter64.asm:1322
> > +macro loadEncodedThisValue(baseValue, dest)
> > +   loadpFromInstruction(6, dest)
> > +   bineq 0, dest, .loadEncodedThisValueDone
> > +   move baseValue, dest
> > +   .loadEncodedThisValueDone:
> > +end
> 
> I think this should be bpneq. bineq only compares the low 32 bits. Also, You
> could move it to LowLevelInterpreter.asm

Nice catch, Thanks!

-- 
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/20160801/13f9b155/attachment.html>


More information about the webkit-unassigned mailing list