It should not impact GC performance. And the patch has only positive impact in my opinion. So +1...
Thanks, xiaofeng On Fri, Apr 3, 2009 at 5:47 PM, Egor Pasko <[email protected]> wrote: > On the 0x588 day of Apache Harmony Tim Ellison wrote: >> 2009/4/3 Egor Pasko <[email protected]>: >>> On the 0x588 day of Apache Harmony Oliver Deakin wrote: >>>> I just ran the concurrent tests with M8 and I see the same tests >>>> failing (some intermittently) along with some others, so I do not >>>> believe these are regressions in M9. Based on these failures not being >>>> regressions, and all other tests passing, I'd be happy to progress >>>> with declaring M9. >>> >>> +1 >> >> Egor, I just tested your HARMONY-6137 patch, and it fixes all these >> concurrent failures for me. >> >> I'll leave it up to your judgment about whether it is safe to apply in >> the closing moments of M9 (but it would be nice to have these >> resolved). > > yey! I did not quite expect it to fix all failures :) > > the patch is trivial (telling GCC that the assembly touches memory, > which it really does!). It may affect (GC) performance, but even if it > does correctness is more important. > > so I am +1 to commit it. > > -- > Egor Pasko > > -- http://people.apache.org/~xli
