On 1 July 2016 at 16:34, Wirth, Allan <awi...@akamai.com> wrote: > Thanks for the feedback. I didn’t find that patch before when I searched, so > apologies for the duplicate submission. > > The proposed fix certainly does seem cleaner and more general. Does it > imply though that this patch is incorrect? It fixes the emulation bug > in my use case, and AFAICT does not introduce new emulation bugs.
Well, it depends what you mean by "incorrect". It's pretty common in dealing with a large and old code base to find good opportunities for small refactorings when you investigate a bug. If we allow bugs to be fixed with the smallest and most expedient change, then problems gradually pile up and the codebase becomes unmaintainable. So we often ask patch submitters to do a bit of cleanup in the process of fixing their bug. In this case, fixing the bug in the way that I suggest will fix it for all architectures, not just x86-64, improve the code by deleting an #ifdef, and remove a trap that will otherwise be waiting for the next new architecture that has support contributed to it. thanks -- PMM
