Re: UPDATE: devel/boehm-gc 8.0.4
On Thu, 2021-02-04 at 15:25 +0100, Janne Johansson wrote: > Den tors 4 feb. 2021 kl 14:47 skrev Janne Johansson : > > > > > > > > > k...@intricatesoftware.com wrote: > > > > > > > > Update to 8.0.4 with the following additional changes: > > > > Additional testing on mips, powerpc, etc. would be appreciated. > > Mips64: > > > > make check-TESTS > > > > > > Testsuite summary for gc 8.0.4 > > > > > > # TOTAL: 17 > > # PASS: 17 > > > > > > w3m still building... > > > w3m on wikipedia and other sites works fine too. > Great. Thank you for testing it. -Kurt
Re: UPDATE: devel/boehm-gc 8.0.4
Den tors 4 feb. 2021 kl 14:47 skrev Janne Johansson : > > k...@intricatesoftware.com wrote: > > > Update to 8.0.4 with the following additional changes: > > > Additional testing on mips, powerpc, etc. would be appreciated. > > Mips64: > > make check-TESTS > > > Testsuite summary for gc 8.0.4 > > > # TOTAL: 17 > # PASS: 17 > > > w3m still building... > w3m on wikipedia and other sites works fine too. -- May the most significant bit of your life be positive.
Re: UPDATE: devel/boehm-gc 8.0.4
On Wed, 2021-02-03 at 22:42 +, Stuart Henderson wrote: > On 2021/02/02 18:40, k...@intricatesoftware.com wrote: > > > > Update to 8.0.4 with the following additional changes: > > - Use mmap via the configure argument instead of hard-coding it. > > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream > > code conventions. > > - Remove unused ELF_CLASS define for OpenBSD. > > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD. > > - Don't use old GC_find_limit_openbsd() which is only needed for > > uthreads. > > > > I've tested this on i386, amd64, sparc64 and aarch64 using the > > port's regress tests and by using w3m. Note that on aarch64 the > > current boehm-gc port isn't stable and w3m will crash it pretty > > easily. This update has not fixed that (aarch64 is still unstable > > but the regress tests pass). > > > > Additional testing on mips, powerpc, etc. would be appreciated. > I'm OK with this, it's a significant clean-up for what is one of the > trickiest ports in the tree, but tests on other arches would be welcome. > > Note to testers: building w3m against the new boehm-gc and browsing > a selection of pages is good for testing; following a few links around > wikipedia or something is often enough to pick up problems. > > Kurt, I know it's just papering over the problem but what do you think > about setting -O1 in the w3m port for aarch64 for now? > Thanks. I'd like a bit more time to investigate w3m on aarch64, but if I can't figure it out, sure we can set it to -O1 so it works. -Kurt
Re: UPDATE: devel/boehm-gc 8.0.4
On Wed, 2021-02-03 at 23:37 -0500, George Koehler wrote:
> On Tue, 2 Feb 2021 18:40:50 -0500 (EST)
> k...@intricatesoftware.com wrote:
>
> >
> > Update to 8.0.4 with the following additional changes:
> > - Use mmap via the configure argument instead of hard-coding it.
> > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream
> > code conventions.
> > - Remove unused ELF_CLASS define for OpenBSD.
> > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD.
> > - Don't use old GC_find_limit_openbsd() which is only needed for
> > uthreads.
> >
> > I've tested this on i386, amd64, sparc64 and aarch64 using the
> > port's regress tests and by using w3m
> Your update to 8.0.4 passes the port's "make test" on powerpc and
> powerpc64. It works with net/clic on powerpc64. ok gkoehler@
>
> This update fixes -fno-common
>
> I like how you adjusted the patches, and you tested the port on
> multiple platforms. I had added the mmap patch along with the
> powerpc64 support, but upstream has changed their mmap code, so you
> needed to adjust the patch. Defining USE_MMAP_ANON for each arch
> is exactly what upstream does for systems like Mac (DARWIN). Another
> way is at include/private/gcconfig.h.orig:2906, where upstream defines
> USE_MMAP_ANON for all arches of LINUX.--George
Thanks for the review and tests. Indeed I missed that part of
gcconfig.h, thanks for pointing it out. My intent was to have
the configure argument control the use of mmap, but I see that
I didn't get that quite right due to this part:
#if defined(USE_MMAP_ANON) && !defined(USE_MMAP)
# define USE_MMAP 1
#elif defined(LINUX) && defined(USE_MMAP)
/* The kernel may do a somewhat better job merging mappings etc.*/
/* with anonymous mappings. */
# define USE_MMAP_ANON
#endif
The way I have it now it always uses mmap, so I'd like to
revise that to remove the USE_MMAP_ANON per arch and adjust
this section above as follows to allow the configure argument
to control the use of mmap:
#if defined(USE_MMAP_ANON) && !defined(USE_MMAP)
# define USE_MMAP 1
#elif defined(LINUX) && defined(USE_MMAP)
/* The kernel may do a somewhat better job merging mappings etc.*/
/* with anonymous mappings. */
# define USE_MMAP_ANON
#elif defined(OPENBSD) && defined(USE_MMAP)
# define USE_MMAP_ANON
#endif
I'm not sure if my MUA will mangle the in-line diff so I'll
attach it was well.
Index: Makefile
===
RCS file: /cvs/ports/devel/boehm-gc/Makefile,v
retrieving revision 1.75
diff -u -p -u -r1.75 Makefile
--- Makefile24 Aug 2020 21:31:13 - 1.75
+++ Makefile4 Feb 2021 14:13:31 -
@@ -3,8 +3,8 @@
COMMENT-main= garbage collection and memory leak detection for C and C++
COMMENT-atomic=access to hardware provided atomic memory operations
-VERSION= 7.6.0
-LIBAO_VERSION= 7.6.6
+VERSION= 8.0.4
+LIBAO_VERSION= 7.6.10
# has various machine-dependent parts; updates definitely need
# tests on more than just amd64.
@@ -12,7 +12,6 @@ DISTNAME= gc-${VERSION}
PKGNAME-atomic=libatomic_ops-${LIBAO_VERSION}
PKGNAME-main= boehm-gc-${VERSION}
-REVISION= 6
DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
libatomic_ops-${LIBAO_VERSION}${EXTRACT_SUFX}
@@ -20,11 +19,11 @@ DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
MULTI_PACKAGES=-main -atomic
DEBUG_PACKAGES=${BUILD_PACKAGES}
-SHARED_LIBS += gc 4.0 # 1.3
-SHARED_LIBS += gccpp 0.0 # 1.3
-SHARED_LIBS += cord2.3 # 1.3
-SHARED_LIBS += atomic_ops 2.0 # 1.3
-SHARED_LIBS += atomic_ops_gpl 2.0 # 1.3
+SHARED_LIBS += gc 5.0 # 5.3
+SHARED_LIBS += gccpp 1.0 # 5.0
+SHARED_LIBS += cord3.0 # 5.0
+SHARED_LIBS += atomic_ops 3.0 # 2.1
+SHARED_LIBS += atomic_ops_gpl 3.0 # 2.2
CATEGORIES=devel
@@ -49,7 +48,9 @@ CONFIGURE_STYLE= gnu
CONFIGURE_ARGS= --with-libatomic-ops=no \
--enable-cplusplus \
- --enable-threads=pthreads
+ --enable-threads=pthreads \
+ --enable-mmap \
+ --enable-static=yes
post-extract:
@perl -pi -e "s,/usr/local,${PREFIX},g" ${WRKSRC}/doc/gc.man
Index: distinfo
===
RCS file: /cvs/ports/devel/boehm-gc/distinfo,v
retrieving revision 1.11
diff -u -p -u -r1.11 distinfo
--- distinfo30 Dec 2018 17:08:09 - 1.11
+++ distinfo4 Feb 2021 14:13:31 -
@@ -1,4 +1,4 @@
-SHA256 (gc-7.6.0.tar.gz) = oUoosRKb6Q5VzW9xEn/8VZThCR1dVBMVKMJM0MA7fZA=
-SHA256 (libatomic_ops-7.6.6.tar.gz) =
mf6rxfVId/MU20+t6xCfCz4dGlSvtrSz37oecH444HQ=
-SIZE (gc-7.6.0.tar.gz) = 751
-SIZE (libatomic_ops-7.6.6.tar.gz) = 498187
+SHA256 (gc-8.0.4.tar.gz) =
Re: UPDATE: devel/boehm-gc 8.0.4
On Thu, 2021-02-04 at 02:06 +0100, Charlene Wendling wrote: > On Tue, 2 Feb 2021 18:40:50 -0500 (EST) > k...@intricatesoftware.com wrote: > > > > > Update to 8.0.4 with the following additional changes: > > - Use mmap via the configure argument instead of hard-coding it. > > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream > > code conventions. > > - Remove unused ELF_CLASS define for OpenBSD. > > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD. > > - Don't use old GC_find_limit_openbsd() which is only needed for > > uthreads. > > > > I've tested this on i386, amd64, sparc64 and aarch64 using the > > port's regress tests and by using w3m. Note that on aarch64 the > > current boehm-gc port isn't stable and w3m will crash it pretty > > easily. This update has not fixed that (aarch64 is still unstable > > but the regress tests pass). > > > > Additional testing on mips, powerpc, etc. would be appreciated. > I tested on macppc: > > - all 17 tests are passing > - w3m's runtime is still fine Thank you for the test report. -Kurt
Re: UPDATE: devel/boehm-gc 8.0.4
> k...@intricatesoftware.com wrote: > > Update to 8.0.4 with the following additional changes: > > Additional testing on mips, powerpc, etc. would be appreciated. > > I tested on macppc: > > - all 17 tests are passing > - w3m's runtime is still fine > > Charlène. > Mips64: make check-TESTS PASS: cordtest PASS: gctest PASS: leaktest PASS: middletest PASS: smashtest PASS: hugetest PASS: realloc_test PASS: staticrootstest PASS: test_atomic_ops PASS: threadleaktest PASS: threadkey_test PASS: subthreadcreate_test PASS: initsecondarythread_test PASS: test_cpp PASS: disclaim_test PASS: disclaim_bench PASS: disclaim_weakmap_test Testsuite summary for gc 8.0.4 # TOTAL: 17 # PASS: 17 # SKIP: 0 # XFAIL: 0 # FAIL: 0 # XPASS: 0 # ERROR: 0 w3m still building... -- May the most significant bit of your life be positive.
Re: UPDATE: devel/boehm-gc 8.0.4
On Tue, 2 Feb 2021 18:40:50 -0500 (EST)
k...@intricatesoftware.com wrote:
> Update to 8.0.4 with the following additional changes:
> - Use mmap via the configure argument instead of hard-coding it.
> - Define USE_MMAP_ANON for all OpenBSD arch to match upstream
> code conventions.
> - Remove unused ELF_CLASS define for OpenBSD.
> - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD.
> - Don't use old GC_find_limit_openbsd() which is only needed for
> uthreads.
>
> I've tested this on i386, amd64, sparc64 and aarch64 using the
> port's regress tests and by using w3m
Your update to 8.0.4 passes the port's "make test" on powerpc and
powerpc64. It works with net/clic on powerpc64. ok gkoehler@
This update fixes -fno-common
I like how you adjusted the patches, and you tested the port on
multiple platforms. I had added the mmap patch along with the
powerpc64 support, but upstream has changed their mmap code, so you
needed to adjust the patch. Defining USE_MMAP_ANON for each arch
is exactly what upstream does for systems like Mac (DARWIN). Another
way is at include/private/gcconfig.h.orig:2906, where upstream defines
USE_MMAP_ANON for all arches of LINUX.--George
> Index: Makefile
> ===
> RCS file: /cvs/ports/devel/boehm-gc/Makefile,v
> retrieving revision 1.75
> diff -u -p -u -r1.75 Makefile
> --- Makefile 24 Aug 2020 21:31:13 - 1.75
> +++ Makefile 2 Feb 2021 23:04:37 -
> @@ -3,8 +3,8 @@
> COMMENT-main=garbage collection and memory leak detection for C and
> C++
> COMMENT-atomic= access to hardware provided atomic memory operations
>
> -VERSION= 7.6.0
> -LIBAO_VERSION= 7.6.6
> +VERSION= 8.0.4
> +LIBAO_VERSION= 7.6.10
> # has various machine-dependent parts; updates definitely need
> # tests on more than just amd64.
>
> @@ -12,7 +12,6 @@ DISTNAME= gc-${VERSION}
>
> PKGNAME-atomic= libatomic_ops-${LIBAO_VERSION}
> PKGNAME-main=boehm-gc-${VERSION}
> -REVISION=6
>
> DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
> libatomic_ops-${LIBAO_VERSION}${EXTRACT_SUFX}
> @@ -20,11 +19,11 @@ DISTFILES=${DISTNAME}${EXTRACT_SUFX} \
> MULTI_PACKAGES= -main -atomic
> DEBUG_PACKAGES= ${BUILD_PACKAGES}
>
> -SHARED_LIBS += gc 4.0 # 1.3
> -SHARED_LIBS += gccpp 0.0 # 1.3
> -SHARED_LIBS += cord2.3 # 1.3
> -SHARED_LIBS += atomic_ops 2.0 # 1.3
> -SHARED_LIBS += atomic_ops_gpl 2.0 # 1.3
> +SHARED_LIBS += gc 5.0 # 5.3
> +SHARED_LIBS += gccpp 1.0 # 5.0
> +SHARED_LIBS += cord3.0 # 5.0
> +SHARED_LIBS += atomic_ops 3.0 # 2.1
> +SHARED_LIBS += atomic_ops_gpl 3.0 # 2.2
>
> CATEGORIES= devel
>
> @@ -49,7 +48,9 @@ CONFIGURE_STYLE= gnu
>
> CONFIGURE_ARGS= --with-libatomic-ops=no \
> --enable-cplusplus \
> - --enable-threads=pthreads
> + --enable-threads=pthreads \
> + --enable-mmap \
> + --enable-static=yes
>
> post-extract:
> @perl -pi -e "s,/usr/local,${PREFIX},g" ${WRKSRC}/doc/gc.man
> Index: distinfo
> ===
> RCS file: /cvs/ports/devel/boehm-gc/distinfo,v
> retrieving revision 1.11
> diff -u -p -u -r1.11 distinfo
> --- distinfo 30 Dec 2018 17:08:09 - 1.11
> +++ distinfo 2 Feb 2021 23:04:37 -
> @@ -1,4 +1,4 @@
> -SHA256 (gc-7.6.0.tar.gz) = oUoosRKb6Q5VzW9xEn/8VZThCR1dVBMVKMJM0MA7fZA=
> -SHA256 (libatomic_ops-7.6.6.tar.gz) =
> mf6rxfVId/MU20+t6xCfCz4dGlSvtrSz37oecH444HQ=
> -SIZE (gc-7.6.0.tar.gz) = 751
> -SIZE (libatomic_ops-7.6.6.tar.gz) = 498187
> +SHA256 (gc-8.0.4.tar.gz) = Q2oN3GexrAsEBbYalnW8qeB1yBVvTevR0G86VsfNKJ0=
> +SHA256 (libatomic_ops-7.6.10.tar.gz) =
> WH7fYIF/Vtrx4as4pLPHKbjoRv9ntPYqYVcYNwjwma8=
> +SIZE (gc-8.0.4.tar.gz) = 1160528
> +SIZE (libatomic_ops-7.6.10.tar.gz) = 503734
> Index: patches/patch-bdw-gc_pc_in
> ===
> RCS file: /cvs/ports/devel/boehm-gc/patches/patch-bdw-gc_pc_in,v
> retrieving revision 1.1
> diff -u -p -u -r1.1 patch-bdw-gc_pc_in
> --- patches/patch-bdw-gc_pc_in9 Dec 2009 20:36:49 - 1.1
> +++ patches/patch-bdw-gc_pc_in2 Feb 2021 23:04:37 -
> @@ -1,10 +1,11 @@
> $OpenBSD: patch-bdw-gc_pc_in,v 1.1 2009/12/09 20:36:49 sthen Exp $
> bdw-gc.pc.in.origMon Dec 7 10:57:34 2009
> -+++ bdw-gc.pc.in Mon Dec 7 10:57:46 2009
> +Index: bdw-gc.pc.in
> +--- bdw-gc.pc.in.orig
> bdw-gc.pc.in
> @@ -6,5 +6,5 @@ includedir=@includedir@
> Name: Boehm-Demers-Weiser Conservative Garbage Collector
> Description: A garbage collector for C and C++
> Version: @PACKAGE_VERSION@
> --Libs: -L${libdir} -lgc
> -+Libs: -L${libdir} -lgc -pthread
> +-Libs: -L${libdir} @ATOMIC_OPS_LIBS@ -lgc
> ++Libs:
Re: UPDATE: devel/boehm-gc 8.0.4
On Tue, 2 Feb 2021 18:40:50 -0500 (EST) k...@intricatesoftware.com wrote: > Update to 8.0.4 with the following additional changes: > - Use mmap via the configure argument instead of hard-coding it. > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream > code conventions. > - Remove unused ELF_CLASS define for OpenBSD. > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD. > - Don't use old GC_find_limit_openbsd() which is only needed for > uthreads. > > I've tested this on i386, amd64, sparc64 and aarch64 using the > port's regress tests and by using w3m. Note that on aarch64 the > current boehm-gc port isn't stable and w3m will crash it pretty > easily. This update has not fixed that (aarch64 is still unstable > but the regress tests pass). > > Additional testing on mips, powerpc, etc. would be appreciated. I tested on macppc: - all 17 tests are passing - w3m's runtime is still fine Charlène.
Re: UPDATE: devel/boehm-gc 8.0.4
On 2021/02/02 18:40, k...@intricatesoftware.com wrote: > Update to 8.0.4 with the following additional changes: > - Use mmap via the configure argument instead of hard-coding it. > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream > code conventions. > - Remove unused ELF_CLASS define for OpenBSD. > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD. > - Don't use old GC_find_limit_openbsd() which is only needed for > uthreads. > > I've tested this on i386, amd64, sparc64 and aarch64 using the > port's regress tests and by using w3m. Note that on aarch64 the > current boehm-gc port isn't stable and w3m will crash it pretty > easily. This update has not fixed that (aarch64 is still unstable > but the regress tests pass). > > Additional testing on mips, powerpc, etc. would be appreciated. I'm OK with this, it's a significant clean-up for what is one of the trickiest ports in the tree, but tests on other arches would be welcome. Note to testers: building w3m against the new boehm-gc and browsing a selection of pages is good for testing; following a few links around wikipedia or something is often enough to pick up problems. Kurt, I know it's just papering over the problem but what do you think about setting -O1 in the w3m port for aarch64 for now?
Re: UPDATE: devel/boehm-gc 8.0.4
On Tue, 2021-02-02 at 18:40 -0500, k...@intricatesoftware.com wrote: > Update to 8.0.4 with the following additional changes: > - Use mmap via the configure argument instead of hard-coding it. > - Define USE_MMAP_ANON for all OpenBSD arch to match upstream > code conventions. > - Remove unused ELF_CLASS define for OpenBSD. > - Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD. > - Don't use old GC_find_limit_openbsd() which is only needed for > uthreads. > > I've tested this on i386, amd64, sparc64 and aarch64 using the > port's regress tests and by using w3m. Note that on aarch64 the > current boehm-gc port isn't stable and w3m will crash it pretty > easily. This update has not fixed that (aarch64 is still unstable > but the regress tests pass). > > Additional testing on mips, powerpc, etc. would be appreciated. Regarding the w3m crashes on aarch64, I've noticed that dropping the optimization level from -O2 to -O1 for w3m stabilizes it for me. However, I don't see the connection with how this interacts with boehm-gc. It looks like clang -O2 adds -vectorize-loops and -vectorize-slp which seem unrelated to things that would affect boehm-gc. -Kurt
UPDATE: devel/boehm-gc 8.0.4
Update to 8.0.4 with the following additional changes:
- Use mmap via the configure argument instead of hard-coding it.
- Define USE_MMAP_ANON for all OpenBSD arch to match upstream
code conventions.
- Remove unused ELF_CLASS define for OpenBSD.
- Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD.
- Don't use old GC_find_limit_openbsd() which is only needed for
uthreads.
I've tested this on i386, amd64, sparc64 and aarch64 using the
port's regress tests and by using w3m. Note that on aarch64 the
current boehm-gc port isn't stable and w3m will crash it pretty
easily. This update has not fixed that (aarch64 is still unstable
but the regress tests pass).
Additional testing on mips, powerpc, etc. would be appreciated.
Index: Makefile
===
RCS file: /cvs/ports/devel/boehm-gc/Makefile,v
retrieving revision 1.75
diff -u -p -u -r1.75 Makefile
--- Makefile24 Aug 2020 21:31:13 - 1.75
+++ Makefile2 Feb 2021 23:04:37 -
@@ -3,8 +3,8 @@
COMMENT-main= garbage collection and memory leak detection for C and C++
COMMENT-atomic=access to hardware provided atomic memory operations
-VERSION= 7.6.0
-LIBAO_VERSION= 7.6.6
+VERSION= 8.0.4
+LIBAO_VERSION= 7.6.10
# has various machine-dependent parts; updates definitely need
# tests on more than just amd64.
@@ -12,7 +12,6 @@ DISTNAME= gc-${VERSION}
PKGNAME-atomic=libatomic_ops-${LIBAO_VERSION}
PKGNAME-main= boehm-gc-${VERSION}
-REVISION= 6
DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
libatomic_ops-${LIBAO_VERSION}${EXTRACT_SUFX}
@@ -20,11 +19,11 @@ DISTFILES= ${DISTNAME}${EXTRACT_SUFX} \
MULTI_PACKAGES=-main -atomic
DEBUG_PACKAGES=${BUILD_PACKAGES}
-SHARED_LIBS += gc 4.0 # 1.3
-SHARED_LIBS += gccpp 0.0 # 1.3
-SHARED_LIBS += cord2.3 # 1.3
-SHARED_LIBS += atomic_ops 2.0 # 1.3
-SHARED_LIBS += atomic_ops_gpl 2.0 # 1.3
+SHARED_LIBS += gc 5.0 # 5.3
+SHARED_LIBS += gccpp 1.0 # 5.0
+SHARED_LIBS += cord3.0 # 5.0
+SHARED_LIBS += atomic_ops 3.0 # 2.1
+SHARED_LIBS += atomic_ops_gpl 3.0 # 2.2
CATEGORIES=devel
@@ -49,7 +48,9 @@ CONFIGURE_STYLE= gnu
CONFIGURE_ARGS= --with-libatomic-ops=no \
--enable-cplusplus \
- --enable-threads=pthreads
+ --enable-threads=pthreads \
+ --enable-mmap \
+ --enable-static=yes
post-extract:
@perl -pi -e "s,/usr/local,${PREFIX},g" ${WRKSRC}/doc/gc.man
Index: distinfo
===
RCS file: /cvs/ports/devel/boehm-gc/distinfo,v
retrieving revision 1.11
diff -u -p -u -r1.11 distinfo
--- distinfo30 Dec 2018 17:08:09 - 1.11
+++ distinfo2 Feb 2021 23:04:37 -
@@ -1,4 +1,4 @@
-SHA256 (gc-7.6.0.tar.gz) = oUoosRKb6Q5VzW9xEn/8VZThCR1dVBMVKMJM0MA7fZA=
-SHA256 (libatomic_ops-7.6.6.tar.gz) =
mf6rxfVId/MU20+t6xCfCz4dGlSvtrSz37oecH444HQ=
-SIZE (gc-7.6.0.tar.gz) = 751
-SIZE (libatomic_ops-7.6.6.tar.gz) = 498187
+SHA256 (gc-8.0.4.tar.gz) = Q2oN3GexrAsEBbYalnW8qeB1yBVvTevR0G86VsfNKJ0=
+SHA256 (libatomic_ops-7.6.10.tar.gz) =
WH7fYIF/Vtrx4as4pLPHKbjoRv9ntPYqYVcYNwjwma8=
+SIZE (gc-8.0.4.tar.gz) = 1160528
+SIZE (libatomic_ops-7.6.10.tar.gz) = 503734
Index: patches/patch-bdw-gc_pc_in
===
RCS file: /cvs/ports/devel/boehm-gc/patches/patch-bdw-gc_pc_in,v
retrieving revision 1.1
diff -u -p -u -r1.1 patch-bdw-gc_pc_in
--- patches/patch-bdw-gc_pc_in 9 Dec 2009 20:36:49 - 1.1
+++ patches/patch-bdw-gc_pc_in 2 Feb 2021 23:04:37 -
@@ -1,10 +1,11 @@
$OpenBSD: patch-bdw-gc_pc_in,v 1.1 2009/12/09 20:36:49 sthen Exp $
bdw-gc.pc.in.orig Mon Dec 7 10:57:34 2009
-+++ bdw-gc.pc.in Mon Dec 7 10:57:46 2009
+Index: bdw-gc.pc.in
+--- bdw-gc.pc.in.orig
bdw-gc.pc.in
@@ -6,5 +6,5 @@ includedir=@includedir@
Name: Boehm-Demers-Weiser Conservative Garbage Collector
Description: A garbage collector for C and C++
Version: @PACKAGE_VERSION@
--Libs: -L${libdir} -lgc
-+Libs: -L${libdir} -lgc -pthread
+-Libs: -L${libdir} @ATOMIC_OPS_LIBS@ -lgc
++Libs: -L${libdir} @ATOMIC_OPS_LIBS@ -lgc -pthread
Cflags: -I${includedir}
Index: patches/patch-dyn_load_c
===
RCS file: patches/patch-dyn_load_c
diff -N patches/patch-dyn_load_c
--- /dev/null 1 Jan 1970 00:00:00 -
+++ patches/patch-dyn_load_c2 Feb 2021 23:04:37 -
@@ -0,0 +1,33 @@
+$OpenBSD$
+
+- Remove old HAVE_DL_ITERATE_PHDR workaround for OpenBSD
+
+Index: dyn_load.c
+--- dyn_load.c.orig
dyn_load.c
+@@ -81,13 +81,6 @@ STATIC GC_has_static_roots_func GC_has_static_roots =
+ # define ELFSIZE ARCH_ELFSIZE
+ #endif
+
+-#if defined(OPENBSD)
+-# include
+-# if (OpenBSD >= 200519) && !defined(HAVE_DL_ITERATE_PHDR)
+-#
