This is originally derived from the pkgsrc-joyent set for openjdk11, considerably modified. Cut from the jdk15 patches as of jdk15+32. See also README-zero.txt for patches to build a project zero variant. Most patches -p0 JDK17: 17.0.10 Drop the make/modules/jdk.net/Lib.gmk patch, as it's guarded in such a way as will never be used on solaris. Need to add illumos-port-18.patch (derived from jdk21's illumos-port-24.patch) so that test/jdk/java/io/File/libGetXSpace.c will compile. 17.0.9 Much rework around safefetch. Removed illumos-port-12.patch, and removed the is_safefetch_fault() block entirely. (Looked at the Windows port, and that block was removed there in this release. It wasn't present at all in other platforms.) Copied the new assembler src/hotspot/os_cpu/linux_x86/safefetch_linux_x86_64.S to src/hotspot/os_cpu/solaris_x86/safefetch_solaris_x86_64.S, see illumos-port-17.patch In os_solaris_x86.cpp, the end of print_context() is now split off into print_tos_pc() We need an implementation of os::can_trim_native_heap() and trim_native_heap(); simply add a stub in os_solaris.inline.hpp to return false like pretty much every other platform except linux does. 17.0.8 Reinstate make/data/charsetmapping/stdcs-solaris, removal broke the build. See illumos-port-16.patch Cleanup: remove TAR_CREATE_FILE_PARAM TAR_CREATE_EXTRA_PARAM Cleanup: Removed another STLPORT reference 17.0.7 Minor patch noise. Removed an STLPORT patch. 17.0.6 Minor patch noise. 17.0.5 A number of functions centralised into os_posix 17.0.4 Build broken by https://www.illumos.org/issues/14418. That did 2 things - (1) exposed memcntl and meminfo by default, and (2) changed the signature for memcntl from caddr_t to void so there's a mismatch. The fix adopted is to modify the internal java signature for memcntl to the new version, which still allows builds on older releases as the old definition in sys/mman.h was effectively invisible there. Remove some unnecessary patches src/java.base/unix/native/libjli/java_md.h - new comment is good src/java.desktop/unix/native/common/awt/fontpath.c - shouldn't need to check for SunOS 5.8/5.9 Remove HS_DTRACE_WORKAROUND_TAIL_CALL_BUG, the bug it works around was fixed in 2008 Remove remaining SUNPRO and related MLIB_NO_LIBSUNMATH checks src/java.desktop/unix/native/common/awt/X11Color.c - the complexity is over 20 years old and no longer relevant make/autoconf/flags-cflags.m4 - adding -DTRIMMED is useless src/java.desktop/unix/native/libawt_xawt/awt/awt_InputMethod.c - unnecessary on illumos and current Solaris (and my S10 system too) use the 64-bit rdtsc variant 17.0.2 illumos-port-15.patch to fix the broken ld check 17-35 RC1, no changes 17-29,17-30,17-31,17-32,17-33,17-34 No changes 17-27,17-28 No changes. Packaged 17-28. 17-26 First from the jdk17 stabilisation repository. Same changes as in the first jdk18 tag. Remove extended file io workaround, it's 32-bit only 17-25 SecurityManager deprecation. Added suppression annotations (the way it was done for Windows), and cleaned up the line number noise. 17-24 Line number noise. Clean up some of the compiler warnings. 17-23 Line number noise. os::os_exception_wrapper signature; last argument is a JavaThread not a Thread. No impact as we don't actually use it. 17-22 Signature of reserve_memory_special and pd_reserve_memory_special has changed to pass the page size. All we do is change the signature as it's a no-op on solaris. 17-21 Line number noise. 17-20 Quite a lot of changes this time around. No more aot, so remove patches for that. src/hotspot/share/cds/classListParser.cpp has moved to src/hotspot/share/cds/classListParser.cpp suspend/resume has been essentially gutted Assembler files are all .S not .s now; and the way it's called has changed, so we need to just forget about forcing the studio assembler Other changes made: Also removed solstudio, which also cleans up C_FLAG_REORDER Cleared out some of the sparc pieces Gutted ucrypto 17-19 .mx.jvmci has gone, we don't need it anyway Boot jdk requirement bumped to 16. This triggers a lot of noisy "SIGSEGV happened inside stack but outside yellow and red zone." warnings during the build, but they appear harmless. 17-18 Don't patch jib-profiles.js. We don't use this, there aren't any AIX entries, so putting ours in is a waste of effort. Remove the nbproject patches; they're either wrong or irrelevant, or both. The boolAtPut flag stuff is no more. The whole JVMFlag machinery has been reworked. Renamed to set_bool. Remove patching of JvmOverrideFiles.gmk when we then disable the patch later. 17-17 No significant changes 17-16 Some patch noise this time. Some of the SPARC-related noise simply removed. 17-15 No changes this time 17-14 Remove patch to c1_LIRAssembler.cpp as it's only for SPARC Maybe lir_pack/lir_unpack could go too? Tidied up Awt2dLibraries.gmk More centralization into os_posix: os::die() os::shutdown() os::abort() Tidy up some of the format noise 17-13 Remove the patches for lib-X11.m4; they're actually for S10, so we don't want them anyway. And remove the patch to XRBackendNative.c related to it which is also an S10 fixup. os::fork_and_exec has been centralized MetaspaceShared::is_in_trampoline_frame() has been removed 17-12 Remove saved_rbp_address_offset() from thread_solaris_x86.hpp There's also evidence of xattr support on some platforms having been added; it's not needed for the build so I've ignored it for now. 17-11 Fix memTracker.hpp; signature changed Don't patch deoptimization.cpp, it was SPARC-specific 17-10 No significant changes 17-9 No significant changes 17-8 Man os:: functions were centralized into os_posix.inline.hpp, so needed to remove our duplicates from os_solaris.inline.hpp and os_solaris.cpp. Mostly the central copy was identical, the only odd one was that the Solaris os::connect had some funky error handling 17-7 CLOCK_MONOTONIC is now just on. Needed to remove os::supports_monotonic_clock() from os_solaris.inline.hpp Remove javaTimeMillis() and javaTimeSystemUTC() and javaTimeNanos() and javaTimeNanos_info() from os_solaris.cpp Parker::park and Parker::unpark no longer compile, park.hpp has been modified so we need to make sure _counter is present in os_solaris.hpp; that makes it compile but it then crashes the jvm in there The posix variant removed the guarantee in PlatformParker; we need to do the same, fix the signature, and provide an implementation. (The PlatformParker piece took a lot of effort.) 17-6 we now have perfMemory_posix.cpp, so perfMemory_solaris.ccp should be removed or the build fails with multiply defined symbols 17-5 make/CompileJavaModules.gmk has mostly been gutted; ignore our patch as it seem to build fine without os.cpp needs to include alloca.h, see illumos-port-11.patch 17-4 src/jdk.hotspot.agent/share/classes/sun/jvm/hotspot/asm/Disassembler.java has been refactored, but the new version appears to do the same thing as we need in a platform-independent manner, so drop our patch 17-3 make/common/Modules.gmk has been completely redone; remove our patch 17-2 More memory stuff; pd_uncommit_memory pd_reserve_memory pd_attempt_reserve_memory_at have changed signatures, with an extra bool argument (that appears to be unused, at least linux and aix don't make any use of it; this appears to be a Mac thing) 17-0/17-1 Quite a lot of refactoring of the toolchain stuff. The page_size rework is really quite massive. The old _page_sizes was an array you manipulated directly. It's now a set. This needs various changes: os::Solaris::is_valid_page_size() is just _page_sizes.contains() Not strictly broken, but easy to fix: in mpss_sanity_check, we can simplify the getpagesizes() stuff as we know we're modern Rework listing of valid page sizes This is a bit ugly because we don't really end up using the new _page_sizes, but instead emulate the old array. Still, I can't find examples of how the new way is supposed to work on any other platform. Also need to implement print_memory_mappings, as a no-op (like AIX is) That's enough to make it compile; it blows up with an arithmetic exception, apparently inside apply_ergo(). In mpss_sanity_check, we need to make sure page_size (which is really a pointer to _large_page_size) is initialized to the largest valid page size. Build: env PATH=/usr/bin:/usr/sbin:/usr/sfw/bin:/usr/gnu/bin bash ./configure \ --enable-unlimited-crypto --with-boot-jdk=/usr/jdk/instances/jdk16 \ --with-native-debug-symbols=none \ --with-toolchain-type=gcc \ --disable-dtrace \ --disable-warnings-as-errors \ --enable-deprecated-ports=yes \ --with-jobs=3 env PATH=/usr/bin:/usr/sbin:/usr/sfw/bin:/usr/gnu/bin gmake all
