Positive. We do not use the API stub except on ia32, we call the runtime on x64 and ARM for callback properties.
On Sun, Dec 20, 2009 at 10:58 AM, Erik Corry <[email protected]> wrote: > You are sure that this only affects IA32? > > 2009/12/20 Kevin Millikin <[email protected]> > >> I agree, it's much better to have this code safe for GC. This should be >> seen as a temporary fix. >> >> >> On Sun, Dec 20, 2009 at 6:04 AM, <[email protected]> wrote: >> >>> Kevin, >>> >>> I think you are fighting a loosing battle here. While this change will >>> certainly >>> fix the reported problem, I think a safer approach would be to not drop >>> the >>> handles when calling ComputeLoadCallback in the first place. >>> >>> LGTM especially if you clarify the confusing comment. >>> >>> -Ivan >>> >>> >>> http://codereview.chromium.org/504071/diff/1/4 >>> File src/ia32/stub-cache-ia32.cc (right): >>> >>> http://codereview.chromium.org/504071/diff/1/4#newcode818 >>> src/ia32/stub-cache-ia32.cc:818<http://codereview.chromium.org/504071/diff/1/4#newcode818src/ia32/stub-cache-ia32.cc:818>: >>> // collection but instead return a >>> failure object. >>> This comment is very confusing. Isn't it so that emitting the code to >>> call PopHandleScope can cause a GC? This is not what I read in your >>> comment here. >>> At the least please clarify that the two instances of the word "call" do >>> not mean the same call. >>> >>> http://codereview.chromium.org/504071 >>> >>> -- >>> v8-dev mailing list >>> [email protected] >>> http://groups.google.com/group/v8-dev >>> >> >> -- >> v8-dev mailing list >> [email protected] >> http://groups.google.com/group/v8-dev >> > > -- > v8-dev mailing list > [email protected] > http://groups.google.com/group/v8-dev > -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
