[Webkit-unassigned] [Bug 99213] [ARMv7] Neither linkCall() nor linkPointer() should flush code

[bugzilla-daemon at webkit.org]

https://bugs.webkit.org/show_bug.cgi?id=99213


Benjamin Poulain <benjamin at webkit.org> changed:

           What    |Removed                     |Added
----------------------------------------------------------------------------
 Attachment #169195|review?                     |
               Flag|                            |




--- Comment #13 from Benjamin Poulain <benjamin at webkit.org>  2012-10-26 19:08:09 PST ---
(From update of attachment 169195)
View in context: https://bugs.webkit.org/attachment.cgi?id=169195&action=review

I remove the review flag for now.
I think this is a nice patch but I would really like it to be completely generic. ARM is not the only platform without completely consistent caches.

I think we should follow Gavin's advices here. Yong, any opinion?

> Source/JavaScriptCore/ChangeLog:10
> +        LinkBuffer doesn't need to flush code during linking. It will
> +        eventually flush the whole executable. Fixing this gives >%5
> +        sunspider boost (on QNX).

After profiling on hardware, I find it strange you get so much improvement from this. It is pretty far from a hotspot.

It is still a good idea to fix this though.

-- 
Configure bugmail: https://bugs.webkit.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are the assignee for the bug.