On Sat, Nov 09, 2024 at 07:23:31PM -0500, George Koehler wrote: > On Sat, 9 Nov 2024 11:50:12 +0100 > Jeremie Courreges-Anglas <j...@wxcvbn.org> wrote: > > > I propose I commit this update, > > and then you add your runtime fix for powerpc64 on top. > > Here's my diff on top, with the powerpc thread-switching fix, and the > powerpc64 runtime fixes.
tl;dr: ok jca@ > Also, "make update-patches" wants to trim > some empty lines from patch*aarch64*. Is this ok? Sorry, this is just me being sloppy. I've synced those in a separate commit so that cvs log doesn't get more confusing. > WRKSRC/gdb/make-init-c was not finding the _initialize_ppc64obsd_* > functions; change their style to what make-init-c expects. Get rid of > inferior_ptid, which had pid 0. Drop 32-bit bfd_arch_rs6000. All of those changes make sense to me. > I also want to report that 32-bit archs (powerpc but not powerpc64) > give lots of this warning, > | /usr/local/include/xxhash.h:617:16: warning: unknown attribute > | 'rpl_malloc' ignored [-Wunknown-attributes] > because <xxhash.h> has "__attribute__((malloc))" but > WRKSRC/gnulib/import/stdlib.h does "#define malloc rpl_malloc", after > WRKSRC/gnulib/import/m4/malloc.m4 checks "whether malloc is ptrdiff_t > safe", which is "no" on 32-bit archs. Ugh, we want to address that at some point I think. Thanks, -- jca
