On Sun, Apr 2, 2023 at 1:07 PM Jeff Law <jeffreya...@gmail.com> wrote: > > > > On 4/2/23 14:06, Andrew Pinski wrote: > > On Sun, Apr 2, 2023 at 12:55 PM Jeff Law via Gcc-patches > > <gcc-patches@gcc.gnu.org> wrote: > >> > >> > >> > >> On 3/31/23 12:44, Sam James wrote: > >>> > >>> Kito Cheng <kito.ch...@gmail.com> writes: > >>> > >>>> It's not the RISC-V part, so this requires a global maintainer there I > >>>> think? > >>>> > >>> > >>> Someone able to look at the system.h bit? It should be trivial, there's > >>> no uses left and it was added purely for a bug like this in the past > >>> (see commit message). > >> You assert that pthread.h is no longer used... But ISTM you really need > >> to go back to when the include was added, understand why it was added > >> and explain why it is no longer needed. > > > > It was needed for the JIT front-end at the time used pthread_mutex_* > > and pthread.h could use a poisoned identifier (I think it was calloc); > > the INCLUDE_PTHREAD_H was added with r13-1350-g49d508065bdd36. The JIT > > front-end moved to using C++11's mutex in r13-4164-g0a62889c7a155f and > > moved away from using pthread.h but didn't remove INCLUDE_PTHREAD_H > > support. > > > > I hope that help explains why it is no longer needed and how it became > > even unused. > SO I'm confused, what does this have to do with RISC-V?
So this was originally submitted with a patch against gcc/config/riscv/genrvv-type-indexer.cc as it was noticed that during the development of r13-6662-g0e6f87835ccabf, that INCLUDE_PTHREAD_H became unused. The original patch against genrvv-type-indexer.cc was to use INCLUDE_PTHREAD_H which was obviously wrong and then I helped sam come up with the correct fix and we found that INCLUDE_PTHREAD_H was not being used any more so Sam submitted a patch to remove it so it would not be used accidently by anyone again. Thanks, Andrew > > jeff
