During an exp-run for llvm 13 (see bug 258209), it turned out that
java/openjdk11 through openjdk13 fail to build with clang 13:
=== Output from failing command(s) repeated here ===
* For target jdk__packages_attribute.done:
These crashes are all caused by the markOop/markOopDesc classes, which
are used to keep track of objects, and which are 'marked' using the low
few bits. (See
https://github.com/openjdk/jdk13u/blob/master/src/hotspot/share/oops/markOop.hpp
).
After some laborious bisecting, I found out that these crashes start
occuring after the upstream commit https://github.com
/llvm/llvm-project/commit/16d03818412 (Return "[CGCall] Annotate this
argument with alignment").
What happens afterwards, is that clang considers the "this" pointer to
always be aligned to the alignment of the actual object, and then
masking or adding a few low bits is not working as expected.
The reason openjdk14 and higher work fine with clang 13, and don't crash
similarly, is that the OpenJDK people completely redid the
markOop/markOopDesc classes in
ae5615c614
("8229258: Rework markOop and markOopDesc into a simpler mark word value
carrier"). E.g, the markOopDesc class was renamed to markWord, and
*stores* a pointer-like value instead of *being* a pointer-like value.
This is a much safer way of handling things.
However, this upstream commit is *very* large, as are a few of its
follow-ups, which is probably the reason why it has not been backported
to JDKs <= 13. I tried manually backporting it, but got lost in many
nasty patch conflicts and problems.
As a workaround, build openjdk8 through 13 with clang12 from the
devel/llvm12 port, for the time being.
In addition, allow openjdk14 through 17 to be built with clang 13, by
adding -Wno-unused-but-set-parameter to the compilation flags.
PR: 258954
Approved by: maintainer timeout (2 weeks)
MFH: 2021Q4
During an exp-run for llvm 12 (see bug 255570), it turned out that
java/openjdk11 does not build with clang 12.0.0:
Creating support/demos/image/jfc/J2Ddemo/J2Ddemo.jar
/usr/local/bin/bash: line 5: 49140 Abort trap (core dumped) /wrkdirs/usr/ports/java/openjdk12/work/openjdk-jdk12u-jdk-12.0.2-10-4/build/bsd-x86_64-server-release/support/interim-image/bin/java -XX:DumpLoadedClassList=/wrkdirs/usr/ports/java/openjdk12/work/openjdk-jdk12u-jdk-12.0.2-10-4/build/bsd-x86_64-server-release/support/link_opt/classlist.raw -Djava.lang.invoke.MethodHandle.TRACE_RESOLVE=true -cp /wrkdirs/usr/ports/java/openjdk12/work/openjdk-jdk12u-jdk-12.0.2-10-4/build/bsd-x86_64-server-release/support/classlist.jar build.tools.classlist.HelloClasslist > /dev/null 2>&1 > /wrkdirs/usr/ports/java/openjdk12/work/openjdk-jdk12u-jdk-12.0.2-10-4/build/bsd-x86_64-server-release/support/link_opt/default_jli_trace.txt
gmake[4]: *** [GenerateLinkOptData.gmk:66: /wrkdirs/usr/ports/java/openjdk12/work/openjdk-jdk12u-jdk-12.0.2-10-4/build/bsd-x86_64-server-release/support/link_opt/classlist] Error 134
This is due to missing backports of upstream commits:
commit c484d8904285652246c3af212a4211b9a8955149
Author: Thomas Stuefe <stuefe@openjdk.org>
Date: Tue Mar 16 05:49:01 2021 +0000
8263557: Possible NULL dereference in Arena::destruct_contents()
Reviewed-by: kbarrett, coleenp
commit 34ae46e2bca691f989d4d84129baf545ff8a7469
Author: Andrew Haley <aph@openjdk.org>
Date: Thu Mar 19 14:53:57 2020 +0000
8241296: Segfault in JNIHandleBlock::oops_do()
Reviewed-by: stefank
Approved by: maintainer timeout (2 weeks)
PR: 255901
MFH: 2021Q2
These changes largely fall into just two categories:
1. Need textproc/gsed for GNU extensions
2. Extraneous escapes that can go away
For #1, there's a further subdivision into those that require autoconf magic
and those that can get away with BINARY_ALIAS=sed=${LOCALBASE}/bin/gsed.
-CURRENT will soon gain GNU extensions, but these will take longer to get to
all supported releases; we must switch them to gsed to ensure we're actually
properly building them as intended.
For #2, I've fixed these as I can and we should upstream these fixes.
PORTREVISION is bumped for all of the above, because we will almost
certainly build these differently when the replacements actually start
working.
These were all detected by the below-referenced exp-run [1]. The patch
included forbids many ordinary characters from being escaped, since we'll
later imbue those with special meanings. This has had the nice side effect
of picking up various things that we didn't handle properly, e.g. \t and \r
for tab and carriage return.
PR: 229925 [1]
Approved by: koobs (mentor)
Approved by: portmgr (blanket: trivial build fixes)
MFH: no (invasive risk)
Differential Revision: https://reviews.freebsd.org/D25185
Some changes have been made in this release that break backwards
compatibility. Let USES=gmake force users to upgrade to this version so
port maintainers don't have to support older versions.
PR: 245725
Exp-run by: antoine
Approved by: portmgr (antoine)
Because of issue with macros in precompiled.hpp, --disable-precompiled-headers is necessary.
Since openjdk compiles for elfv1 by default on big-endian ppc64, use a patch to compile for elfv2.
PR: 243183
Approved by: glewis (maintainer)
* Having /proc mounted is not generally critical, but there is at least one
piece of functionality that does rely on its presence at the moment
(e.g. OperatingSystemMXBean)
* Make sure has_mfdscr() returns false on these machines as using it on
FreeBSD 12.x will cause problems (e.g. SIGILL).
PR: 239368
Sponsored by: The FreeBSD Foundation
* Remove portions of pkg-message that mention needing /proc mounted.
This hasn't been the case for a while afaik (I don't have it mounted
and haven't observed any related problems).
* Sort LIB_DEPENDS and USE_XORG.
* Add a dependency in alsa-lib now that sound is supported on FreeBSD.
* Don't disable dtrace now that it is supported on FreeBSD.
* Less intrusive iconv change.
