Daniel Gredler [Wed, 12 Nov 2025 14:42:30 +0000 (14:42 +0000)]
[PATCH] 8371066: Remove unused class TextSourceLabel and associated class hierarchy This patch removes unused classes, it is added so that font memory leak fix applies cleanly.
Phil Race [Mon, 23 Feb 2026 18:36:00 +0000 (18:36 +0000)]
[PATCH] 8337853: Remove SunLayoutEngineKey and SunLayoutEngineFactory and its cache. Remove layout engine cache that causes temporary directory to fill up with the temporary font files.
Vladimir Petko [Mon, 6 Jul 2026 21:58:32 +0000 (09:58 +1200)]
[PATCH] fix(TemplateTable): use float operations for fast_xaccess() and fstore(). This PR changes fstore and fast_xaccess operations to generate float (4 byte) instructions on S390x. This resolves out of bound read and writes on float that causes a segmentation fault with Compact Object Headers enabled.
Vladimir Petko [Wed, 3 Jun 2026 06:35:57 +0000 (06:35 +0000)]
[PATCH] 8385738: Javadoc does not produce reproducible output due to the snippet ids The top level type generation order is not deterministic due to the directory listing being non-deterministic. Snippet ids are monotonically incremented during the javadoc run. This means that the generated snippet ids are not reproducible. This PR changes Set to the Map so that each page uses a separate snippet id counter.
Bug: https://bugs.openjdk.org/browse/JDK-8385738 Reviewed-by: nbenalla
Gbp-Pq: Name jdk-8385738.diff
FileStore tests fail in container infrastructure due to JDK-8166162.
This patch disables individual test cases requiring FileStore.
The patch is not needed upstream as the bug only occurs in specific
container configuration.
x86 hotspot aseembler incorrectly returns next instruction's PC
for rep stos %eax,%es:(%edi) causing an infinite loop.
This patch removes UnsafeGuard for i386, so that VM crashes
instead of hanging indefinitely.
Patch should not be forwarded as it reverses JDK-8320886 for
i386.
Vladimir Petko [Thu, 24 Sep 2026 07:52:19 +0000 (19:52 +1200)]
print warning to install openjdk-*-jre package
Forwarded: not-needed
fontmanager support was moved to recommends, warn user to
either install -jre package or recommended packages.
Forwarding is not needed as this is a Debian-only issue.
OpenJDK Team [Thu, 24 Sep 2026 07:52:19 +0000 (19:52 +1200)]
Misalign address in the test rather than structure
mk68-support.diff aligned metaspace structures by 4 bytes breaking the test.
Add 1 to address type rather than aligned structure to produce misaligned pointer.
The cause is the use of dh_strip_nondeterminism late in the build
process. This reorganises the jmod files, which in turn changes their
SHA256 checksums. This would not be a problem, except that the
checksums are saved in java.base.jmod *before* the use of
dh_strip_nondeterminism. Performing this stripping immediately after
each jmod file is created results in the checksums being consistent
throughout.
* OpenJDK 25.0.5 snapshot, build 7.
* Cherry-pick upstream patches to resolve FFM font rendering crash
and memory leak (LP: #2155650):
- d/p/jdk-8337853.diff: Temporary directory was filling up
with the temporary font files.
- d/p/jdk-8371066.diff: Remove unused classes so that other font patches
apply cleanly.
- d/p/jdk-8378603.diff: libfontmanager FFM was using mismatched free
and malloc, when overriden with LD_LIBRARY_PRELOAD.
- d/p/jdk-8378607.diff: GlyphLayout cache prevented Fonts from being
garbage collected.
- d/p/jdk-8378615.diff: FFM call stub was holding global JNI reference
to the font.
* d/rules: add --with-external-symbols-in-bundles=full configure
option (JDK-8373246).
* d/B-jre-*.overrides.in: Drop exit-in-shared-library lintian
override.
Vladimir Petko [Mon, 6 Jul 2026 21:58:32 +0000 (09:58 +1200)]
[PATCH] fix(TemplateTable): use float operations for fast_xaccess() and fstore(). This PR changes fstore and fast_xaccess operations to generate float (4 byte) instructions on S390x. This resolves out of bound read and writes on float that causes a segmentation fault with Compact Object Headers enabled.
To accommodate systems like Ubuntu 25.10 that use Rust coreutils,
this PR updates tests that previously assumed busybox was the only
environment to use symlinks for core utilities.
- java/lang/ProcessBuilder/Basic.java: The test is updated to simply
verify that /bin/true and /bin/false are symlinks,
removing the hardcoded check for a /bin/busybox target.
- java/lang/ProcessHandle/InfoTest.java: The test logic is relaxed.
It now confirms that /bin/sleep is a symlink and then uses the
symlink's target as the expected executable name.
Last-Update: 2025-06-19
Gbp-Pq: Name jdk-8359735.patch
The generated image contains libjpackageapplauncheraux.so that
contains a destructor function dcon(). It calls already disposed
logger, causing a crash.
Last-Update: 2024-07-18
Gbp-Pq: Name jdk-8312488.patch
time_t transition in Debian/Ubuntu left 32 bit time_t symbols in glibc.
Looking up 'futimens' via dlsym returns 32 bit version of the function.
This is causing failure to set last modified time
(e.g. instead of year 2017 we get 1976 in the test).
Using the function directly correctly calls 64 bit versions.
When we lookup functions with time_t arguments through dlsym()
calls we should use 64 bit versions.
Last-Update: 2024-07-17
Gbp-Pq: Name jdk-8336529-proposed.patch
i386 declaration of _SafeFetch32_impl contains an underscore
that causes an undefined symbol in libjvm.so.
Disable -Wl,--allow-shlib-undefined link flag that allowed the
build to succeed.
binutils 2.39 changed the signature of init_disassemble_info,
breaking the build.
See https://lore.kernel.org/bpf/20220622231624.t63bkmkzphqvh3kx@alap3.anarazel.de/T/
Vladimir Petko [Wed, 19 Aug 2026 17:55:58 +0000 (19:55 +0200)]
print warning to install openjdk-*-jre package
Forwarded: not-needed
fontmanager support was moved to recommends, warn user to
either install -jre package or recommended packages.
Forwarding is not needed as this is a Debian-only issue.
OpenJDK Team [Wed, 19 Aug 2026 17:55:58 +0000 (19:55 +0200)]
Misalign address in the test rather than structure
mk68-support.diff aligned metaspace structures by 4 bytes breaking the test.
Add 1 to address type rather than aligned structure to produce misaligned pointer.
The cause is the use of dh_strip_nondeterminism late in the build
process. This reorganises the jmod files, which in turn changes their
SHA256 checksums. This would not be a problem, except that the
checksums are saved in java.base.jmod *before* the use of
dh_strip_nondeterminism. Performing this stripping immediately after
each jmod file is created results in the checksums being consistent
throughout.
Vladimir Petko [Mon, 6 Jul 2026 21:58:32 +0000 (09:58 +1200)]
[PATCH] fix(TemplateTable): use float operations for fast_xaccess() and fstore(). This PR changes fstore and fast_xaccess operations to generate float (4 byte) instructions on S390x. This resolves out of bound read and writes on float that causes a segmentation fault with Compact Object Headers enabled.
Vladimir Petko [Wed, 3 Jun 2026 06:35:57 +0000 (06:35 +0000)]
[PATCH] 8385738: Javadoc does not produce reproducible output due to the snippet ids The top level type generation order is not deterministic due to the directory listing being non-deterministic. Snippet ids are monotonically incremented during the javadoc run. This means that the generated snippet ids are not reproducible. This PR changes Set to the Map so that each page uses a separate snippet id counter.
Bug: https://bugs.openjdk.org/browse/JDK-8385738 Reviewed-by: nbenalla
Gbp-Pq: Name jdk-8385738.diff
projects / openjdk-25.git / log