Got a report that build is also flapping on 14.3, so extent the range
of OSVERSIONs.
Allow parallel jobs, but limited to the number of jobs that is
currently used (and works) on the official FreeBSD pkg build cluster.
PR: 290148
Similar to the fix in openjdk17.
For more information and lessons learned see ports commit
fa96c62b2e3cacce280053fda46dedd8b84a1018 and the follow-up commits.
Prevents:
Creating support/modules_libs/java.base/server/libjvm.so from 989 file(s)
=>> Killing runaway build after 7200 seconds with no output
The cmdlines output can also help in debugging the flapping builds if a
race condition is the cause of the 'nm' error. (see previous commits)
Reported-by: https://lists.freebsd.org/archives/freebsd-pkg-fallout/2025-September/807751.html
Plus, use --with-num-cores instead of --with-jobs. With-num-cores still
lets configure override (reduce) the number of parallel jobs if the
memory size is limited. With-jobs sets an absolute number.
See {WRKSRC}/doc/building.md for more information.
Reported-by: https://lists.freebsd.org/archives/freebsd-java/2025-September/003314.html
Bump portrevision so future failures are easy to trace back to specific
port version.
Put back _MAKE_JOBS with a description why it is set empty.
Fixes:
===> Building for openjdk17-17.0.16+8.1_2
Error: 'make -jN' is not supported, use 'make JOBS=N'
NB: this error came from work/jdk17u-jdk-17.0.16-8-1/make/InitSupport.gmk:119
Bump portrevision so future failures are easy to trace back to specific
port version.
Reported-by: https://lists.freebsd.org/archives/freebsd-ports/2025-September/008453.html
The configure script detects cores and memory and calculates itself how many jobs it uses.
checking for number of cores... 28
checking for memory size... 130926 MB
checking for appropriate number of jobs to run in parallel... 28
Override this by the variables of our ports framework.
If this fixes the flapping builds I want to apply it to the other openjdk* ports.
Bump portrevision so future failures are easy to trace back.
Suggested in: https://lists.freebsd.org/archives/freebsd-java/2025-September/003304.html
Build is flapping.
The error that pops up is similar to this, but the filename can change:
nm: 'abstractCompiler.o': Invalid argument
Sometimes the build runs fine and I can't reproduce the failure locally
on a 4-CPU RPI4.
If this stabalizes the build for openjdk17 I will apply it to other
openjdk* ports too.
Mind that openjdk23 and -24 already have this change and don't seem to fail.
Bumped portrevision so I can easily see if future failures use this change
or not.
Reported by: https://portsfallout.com/fallout?port=java%2Fopenjdk17%24
Reviewed by: https://lists.freebsd.org/archives/freebsd-java/2025-September/003296.html
Implement getCommittedVirtualMemorySize() directly using sysctl
kern.proc.vmmap
Implement getOpenFileDescriptorCount() directly using sysctl
kern.proc.nfds
Note on openjdk17 use closefrom where possible (note this use case is
only in openjdk 17, 21, 23 and 24)
Remove the message about the use of fdescfs(5) as this is not needed
for sure, keep the information about procfs(5) as I have not analysed
enough the source code to make sure it is not used anywhere else, but
I don't think it is.
OpenJDK installation inform that fdescfs(5) and procfs(5) should be
mounted when using OpenJDK.
After doing some test and some traces, it seems that this should be
no so usefull, some applications will require it, but most of java
applications doesn't seems to suffer to work or performance when
such pseudo FS is not mounted.
So this change request permit to let local administrator to decide if
it is really needed or not.
Sponsored by: Entersekt (previously Modirum MDpay)
Sponsored by: Klara Inc.
Approved by: 0mp (mentor)
PR: 283364
Generally disable warnings-as-errors for all openjdk ports, removing all
the conditionals we have now for particular architectures and/or
compilers. It does not make sense to have this on by default, unless you
are an upstream maintainer.
PR: 281188
Approved by: maintainer timeout (2 weeks)
MFH: 2024Q3
Clang 19 has a number of new warnings, and because java/openjdk17
compiles with -Werror by default, it results in errors similar to:
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.12-7-1/src/hotspot/os/posix/signals_posix.cpp:1681:20: error: cast from 'void (*)(int, siginfo_t *, ucontext_t *)' (aka 'void (*)(int, __siginfo *, __ucontext *)') to 'void (*)(int)' converts to incompatible function type [-Werror,-Wcast-function-type-mismatch]
1681 | act.sa_handler = (void (*)(int)) SR_handler;
| ^~~~~~~~~~~~~~~~~~~~~~~~~~
I propose to disable warnings-as-errors for this port, similar to what
is already done for openjdk8 and openjdk11. It does not make sense to
have this on by default, unless you are an upstream maintainer.
PR: 280561
Approved by: glewis (maintainer)
MFH: 2024Q3
Thanks again to jbeich for noticing various things I overlooked.
Reviewed by: acm, ashish, eduardo, ehaupt, eugen, jbeich, martymac, mat, nobukata, sunpoet, tagattie, yuri, arrowd, kde
Differential Revision: https://reviews.freebsd.org/D45141
Similar to certificate management with certctl(8) users can use custom
enterprise CAs. These need to be assembled into a custom cacerts file.
Therefore, an admin can install a custom truststore to an arbitrary
location and have the OpenJDK port symlink to it by adding
OPENJDK_SYMLINK_CACERTS to make.conf.
E.g., OPENJDK_SYMLINK_CACERTS=/usr/local/etc/ssl/cacerts.
PR: 229329
Approved by: jrm (mentor), otis (mentor), glewis
Differential Revision: https://reviews.freebsd.org/D42720
Backport the patch from openjdk19 to fix the recently appearing error:
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.9-9-1/src/hotspot/os_cpu/bsd_ppc/os_bsd_ppc.cpp:194:29: error: no member named 'is_safefetch_fault' in 'StubRoutines'
if (pc && StubRoutines::is_safefetch_fault(pc)) {
~~~~~~~~~~~~~~^
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.9-9-1/src/hotspot/os_cpu/bsd_ppc/os_bsd_ppc.cpp:195:52: error: no member named 'continuation_for_safefetch_fault' in 'StubRoutines'
os::Posix::ucontext_set_pc(uc, StubRoutines::continuation_for_safefetch_fault(pc));
~~~~~~~~~~~~~~^
Building java/openjdk17 with lld 17 results in the following link errors:
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZ21WB_HandshakeWalkStackE16TraceSelfClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZ26WB_AsyncHandshakeWalkStackE16TraceSelfClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN12JvmtiEnvBase27check_for_periodic_clean_upEvE28ThreadInsideIterationClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN15G1RemSetSummary6updateEvE11CollectData' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN16SATBMarkQueueSet22set_active_all_threadsEbbE22SetThreadActiveClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN16SATBMarkQueueSet23abandon_partial_markingEvE25AbandonThreadQueueClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN18G1ConcurrentRefine30get_and_reset_refinement_statsEvE12CollectStats' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN19G1DirtyCardQueueSet12abandon_logsEvE23AbandonThreadLogClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN19G1DirtyCardQueueSet16concatenate_logsEvE27ConcatenateThreadLogClosure' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN19G1DirtyCardQueueSet30get_and_reset_refinement_statsEvE12CollectStats' failed: symbol not defined
ld: error: version script assignment of 'SUNWprivate_1.1' to symbol '_ZTVZN7Threads25change_thread_claim_tokenEvE11ResetClaims' failed: symbol not defined
c++: error: linker command failed with exit code 1 (use -v to see invocation)
This is because lld 17 defaults to errors when undefined symbols are
referenced in linker version scripts. The problem is due to the Makefile
JvmMapfile.gmk, which generates a linker version script. It uses "nm
--defined-only" to dump symbols in object files, but this also includes
local (hidden) symbols. Add "--extern-only" to make it only dump global
symbols.
PR: 274109
Approved by: maintainer timeout (10 days)
MFH: 2023Q4
* Disable CompressedClassPointers on aarch64. This fixes the build
on some aarch64 machines where it previously failed, e.g. RPI4,
and doesn't seem to affect the build on others (AWS aarch64 hardware).
This is based on and uses a patch initially created by ronald@
PR: 260187
Tested by: ronald@
During an exp-run for llvm 15 (see bug 265425), it turned out that
java/openjdk17 failed to build with clang 15:
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.5-8-1/src/java.base/unix/native/libnet/DefaultProxySelector.c:374:41: error: passing arguments to a function without a prototype is deprecated in all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
proxies = (*g_proxy_resolver_lookup)(resolver, uri, NULL, &error);
^
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.5-8-1/src/java.base/unix/native/libnet/DefaultProxySelector.c:393:63: error: passing arguments to a function without a prototype is deprecated in all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
(*g_network_address_parse_uri)(proxies[i], 0,
^
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.5-8-1/src/java.base/unix/native/libnet/DefaultProxySelector.c:398:70: error: passing arguments to a function without a prototype is deprecated in all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
phost = (*g_network_address_get_hostname)(conn);
^
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.5-8-1/src/java.base/unix/native/libnet/DefaultProxySelector.c:399:66: error: passing arguments to a function without a prototype is deprecated in all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
pport = (*g_network_address_get_port)(conn);
^
/wrkdirs/usr/ports/java/openjdk17/work/jdk17u-jdk-17.0.5-8-1/src/java.base/unix/native/libnet/DefaultProxySelector.c:435:22: error: passing arguments to a function without a prototype is deprecated in all versions of C and is not supported in C2x [-Werror,-Wdeprecated-non-prototype]
(*g_strfreev)(proxies);
^
This is because clang 15 warns by default about functions without
arguments in their prototypes. Fix this by suppressing the warning.
PR: 268366
Approved by: portmgr (tcberner)
MFH: 2022Q4
Commit b7f05445c00f has added WWW entries to port Makefiles based on
WWW: lines in pkg-descr files.
This commit removes the WWW: lines of moved-over URLs from these
pkg-descr files.
Approved by: portmgr (tcberner)
It has been common practice to have one or more URLs at the end of the
ports' pkg-descr files, one per line and prefixed with "WWW:". These
URLs should point at a project website or other relevant resources.
Access to these URLs required processing of the pkg-descr files, and
they have often become stale over time. If more than one such URL was
present in a pkg-descr file, only the first one was tarnsfered into
the port INDEX, but for many ports only the last line did contain the
port specific URL to further information.
There have been several proposals to make a project URL available as
a macro in the ports' Makefiles, over time.
This commit implements such a proposal and moves one of the WWW: entries
of each pkg-descr file into the respective port's Makefile. A heuristic
attempts to identify the most relevant URL in case there is more than
one WWW: entry in some pkg-descr file. URLs that are not moved into the
Makefile are prefixed with "See also:" instead of "WWW:" in the pkg-descr
files in order to preserve them.
There are 1256 ports that had no WWW: entries in pkg-descr files. These
ports will not be touched in this commit.
The portlint port has been adjusted to expect a WWW entry in each port
Makefile, and to flag any remaining "WWW:" lines in pkg-descr files as
deprecated.
Approved by: portmgr (tcberner)
Adjust ports depending on expired versions of OpenJDK:
- biology/snpeff: 12+ -> 17+
- devel/RStudio: 12 -> 11
- www/closure-compiler: 13 -> 11+
Allow java/openjdk17 to use java/openjdk17-bootstrap on i386
too, this builds just fine on 13.1-i386
Remove jdk12-doc as it is no longer relevant.
Clean up Java version calculations in bsd.java.mk
Reviewed by: jwb, java (glewis)
Differential Revision: https://reviews.freebsd.org/D35280
During an exp-run for llvm 14 (see bug 261742), it turned out that
java/openjdk13 and java/openjdk17 fail to build with clang 14 (but this
also affects openjdk14 through 16):
=== Output from failing command(s) repeated here ===
* For target hotspot_variant-server_libjvm_objs_serviceThread.o:
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
while (((sensors_changed = LowMemoryDetector::has_pending_requests()) |
~^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: note: cast one or both operands to int to silence this warning
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
while (((sensors_changed = LowMemoryDetector::has_pending_requests()) |
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: note: cast one or both operands to int to silence this warning
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
while (((sensors_changed = LowMemoryDetector::has_pending_requests()) |
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: note: cast one or both operands to int to silence this warning
/wrkdirs/usr/ports/java/openjdk13/work/jdk13u-jdk-13.0.10-5-1/src/hotspot/share/runtime/serviceThread.cpp:133:15: error: use of bitwise '|' with boolean operands [-Werror,-Wbitwise-instead-of-logical]
while (((sensors_changed = LowMemoryDetector::has_pending_requests()) |
^~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
... (rest of output omitted)
Although the warning is normally an indication of a potential problem,
in this case a comment just before the affected code explictly mentions
the reason for using bitwise '|' instead of logical '||':
// Process all available work on each (outer) iteration, rather than
// only the first recognized bit of work, to avoid frequently true early
// tests from potentially starving later work. Hence the use of
// arithmetic-or to combine results; we don't want short-circuiting.
(See
<https://github.com/openjdk/jdk/blob/master/src/hotspot/share/runtime/serviceThread.cpp#L140>)
Therefore, we should suppress -Wbitwise-instead-of-logical for clang 14
and higher.
PR: 262845
Approved by: portmgr (build fix blanket)
MFH: 2022Q2
