[Webkit-unassigned] [Bug 12535] Stack-optimizing compilers can trick GC into freeing in-use objects

bugzilla-daemon at webkit.org bugzilla-daemon at webkit.org
Sat Feb 10 22:19:25 PST 2007


http://bugs.webkit.org/show_bug.cgi?id=12535





------- Comment #23 from ian.eng.webkit at gmail.com  2007-02-10 22:19 PDT -------
(In reply to comment #21)
> I measured and this patch causes a 2% performance regression on the iBench, at
> least on Mac OS X (where it is not even needed since the compiler won't
> optimize out object pointers like that). I think this needs to be addressed
> more along the lines of Anrong Hu's approach, where we keep the compiler from
> dropping the pointer. I think copying to alloca()'d memory, or even just
> keeping it in a struct on the stack might do it, and something that works for
> current MSVC seems good enough.
> 

I am fine with Anrong's approach, but that won't cure the problem. What about
another solution: fixing classes that return the address/a reference to a
field?
Like I mentioned in Comment #18, a method can return the value instead of a
reference/address of the field.


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



More information about the webkit-unassigned mailing list