OK, LGTM with the allocate/clone change. On Mon, Mar 8, 2010 at 9:38 AM, <[email protected]> wrote:
> > http://codereview.chromium.org/661179/diff/2001/2002 > File src/ia32/codegen-ia32.cc (right): > > http://codereview.chromium.org/661179/diff/2001/2002#newcode6081 > src/ia32/codegen-ia32.cc:6081: frame_->Spill(result.reg()); > On 2010/03/08 08:26:07, Kevin Millikin wrote: > >> frame() >> > > Done. > > > http://codereview.chromium.org/661179/diff/2001/2002#newcode6105 > src/ia32/codegen-ia32.cc:6105: Result scratch = allocator()->Allocate(); > On 2010/03/08 08:26:07, Kevin Millikin wrote: > >> Allocating a register has a virtual frame effect so the cloned frame >> > and the > >> frame actually reaching the call to AllocateHeapNumber are not >> > identical. > > The register needs to be allocated before the frame is cloned. >> > > Done. > > > http://codereview.chromium.org/661179/diff/2001/2002#newcode6117 > src/ia32/codegen-ia32.cc:6117: result = > frame_->CallRuntime(Runtime::kMath_sqrt, 1); > On 2010/03/08 08:26:07, Kevin Millikin wrote: > >> frame() >> > > Done. > > > http://codereview.chromium.org/661179/diff/2001/2002#newcode6120 > src/ia32/codegen-ia32.cc:6120: frame_->Push(&result); > On 2010/03/08 08:26:07, Kevin Millikin wrote: > >> frame() >> > > Done. > > > http://codereview.chromium.org/661179/diff/2001/2002#newcode6122 > src/ia32/codegen-ia32.cc:6122: Result result = > frame()->CallRuntime(Runtime::kMath_sqrt, 1); > On 2010/03/08 08:26:07, Kevin Millikin wrote: > >> As before (though this one probably fits on a page), it might read a >> > bit better > >> if the small branch of the if...else is first. >> > > Done. > > > http://codereview.chromium.org/661179 > -- v8-dev mailing list [email protected] http://groups.google.com/group/v8-dev
