On Sun, Jan 03 2021, Jeremy Evans <jeremy@openbsd.org> wrote:
> On 01/03 05:26, Jeremie Courreges-Anglas wrote:
>> On Fri, Dec 25 2020, Jeremy Evans <jeremy@openbsd.org> wrote:
>> > Here's a port for the recently released Ruby 3.0.0.
>> >
>> > Release notes at:
>> > http://www.ruby-lang.org/en/news/2020/12/25/ruby-3-0-0-released/
>> >
>> > Tested on amd64. OKs to import?
>>
>> sparc64 strikes again!
>> ruby now tries to use __thread, which fails with our base-gcc:
>>
>> --8<--
>> In file included from vm_core.h:85,
>> from eval_intern.h:5,
>> from vm_trace.c:24:
>> thread_pthread.h: At top level:
>> thread_pthread.h:108: error: thread-local storage not supported for this target
>> In file included from ./include/ruby/internal/arithmetic/char.h:23,
>> from ./include/ruby/internal/arithmetic.h:23,
>> from ./include/ruby/ruby.h:25,
>> from gc.h:3,
>> from vm_dump.c:32:
>> ./include/ruby/internal/arithmetic/int.h: In function 'rb_int2num_inline':
>> ./include/ruby/internal/arithmetic/int.h:146: warning: comparison is always true due to limited range of data type
>> ./include/ruby/internal/arithmetic/int.h:146: warning: comparison is always true due to limited range of data type
>> ./include/ruby/internal/arithmetic/int.h: In function 'rb_uint2num_inline':
>> ./include/ruby/internal/arithmetic/int.h:155: warning: comparison is always true due to limited range of data type
>> In file included from vm_core.h:85,
>> from vm_sync.c:1:
>> thread_pthread.h: At top level:
>> thread_pthread.h:108: error: thread-local storage not supported for this target
>> In file included from vm_core.h:85,
>> from iseq.h:14,
>> from vm_dump.c:36:
>> thread_pthread.h: At top level:
>> thread_pthread.h:108: error: thread-local storage not supported for this target
>> *** Error 1 in /usr/ports/pobj/ruby-3.0.0/ruby-3.0.0 (Makefile:425 'vm_dump.o')
>> *** Error 1 (Makefile:425 'vm_sync.o')
>> *** Error 1 (Makefile:425 'vm_trace.o')
>> *** Error 1 in target 'vm_trace.o'
>> *** Error 1 in target 'vm_sync.o'
>> *** Error 1 in target 'vm_dump.o'
>> *** Error 2 in . (/usr/ports/infrastructure/mk/bsd.port.mk:2925 '/usr/ports/pobj/ruby-3.0.0/.build_done': @cd /usr/ports/pobj/ruby-3.0.0/rub...)
>> *** Error 2 in /usr/ports/lang/ruby/3.0 (/usr/ports/infrastructure/mk/bsd.port.mk:2584 'all': @lock=ruby-3.0.0; export _LOCKS_HELD=" ruby-3...)
>> billy /usr/ports/lang/ruby/3.0$
>> -->8--
>>
>> The attached tarball contains a patch to use the no-__thread fallback,
>> so that we can keep using base-gcc. With this, ruby30 packages on
>> sparc64. ports-gcc supports __thread emulation (like base-clang), but
>> then I'm not sure which compiler MJIT should use.
>>
>> The diff between configure log of 2.7 and 3.0 seems sane, except maybe
>> for
>>
>> -checking whether -D_FORTIFY_SOURCE=2 is accepted as CFLAGS... yes
>> +checking whether -O3 -D_FORTIFY_SOURCE=2 is accepted as CFLAGS... yes
>>
>> Maybe not a new problem, configure log for 2.7 already shows
>>
>> * optflags: -O3
>>
>> I would suggest importing this and fixing problems as they come, but
>> I'm far from a ruby expert.
>
> jca,
>
> I imported ruby 3.0 already, can you send a diff?
d'oh! I had started testing it after the import. :-/
Here's the diff, it should only affect base-gcc where ruby-3.0 doesn't
build. ok?
Index: patches/patch-thread_pthread_h
===================================================================
RCS file: patches/patch-thread_pthread_h
diff -N patches/patch-thread_pthread_h
--- /dev/null 1 Jan 1970 00:00:00 -0000
+++ patches/patch-thread_pthread_h 3 Jan 2021 15:11:15 -0000
@@ -0,0 +1,16 @@
+$OpenBSD$
+
+base-gcc doesn't support __thread
+
+Index: thread_pthread.h
+--- thread_pthread.h.orig
++++ thread_pthread.h
+@@ -72,7 +72,7 @@ typedef struct rb_global_vm_lock_struct {
+
+ #if __STDC_VERSION__ >= 201112
+ #define RB_THREAD_LOCAL_SPECIFIER _Thread_local
+-#elif defined(__GNUC__)
++#elif defined(__clang__)
+ /* note that ICC (linux) and Clang are covered by __GNUC__ */
+ #define RB_THREAD_LOCAL_SPECIFIER __thread
+ #else
--
jca | PGP : 0x1524E7EE / 5135 92C1 AD36 5293 2BDF DDCC 0DFA 74AE 1524 E7EE
No comments:
Post a Comment