In both LLVM builtin dumps you posted, I noticed that EphemeronKeyBarrier is significantly shorter for LLVM 80 bytes vs 348 for the original.
It looks like LLVM has optimised away several conditionals including the check for fp_mode instead assuming it's kSaveFPRegs. Since fp_mode is a parameter to the builtin this does not seem like a safe optimisation. On Tue, 14 Jan 2020 at 09:49, <manjian2...@gmail.com> wrote: > This code reenable root region rematerialize, the last one has a problem > and disable rematerialize. > > I have not yet get my cell phone back, so it has not been tested. > > -- > -- > v8-dev mailing list > v8-dev@googlegroups.com > http://groups.google.com/group/v8-dev > --- > You received this message because you are subscribed to the Google Groups > "v8-dev" group. > To unsubscribe from this group and stop receiving emails from it, send an > email to v8-dev+unsubscr...@googlegroups.com. > To view this discussion on the web visit > https://groups.google.com/d/msgid/v8-dev/c93b4e21-d892-4f08-9414-dc67ff45cb1c%40googlegroups.com > <https://groups.google.com/d/msgid/v8-dev/c93b4e21-d892-4f08-9414-dc67ff45cb1c%40googlegroups.com?utm_medium=email&utm_source=footer> > . > -- -- v8-dev mailing list v8-dev@googlegroups.com http://groups.google.com/group/v8-dev --- You received this message because you are subscribed to the Google Groups "v8-dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to v8-dev+unsubscr...@googlegroups.com. To view this discussion on the web visit https://groups.google.com/d/msgid/v8-dev/CALH_77vMpOmXRsf01uUSH337J%2BPUi-7eqL_qzjZ2iEtNBnSjoQ%40mail.gmail.com.