Created attachment 224730 [details] v0 The bootstrap is on freefall: /home/mikael/bootstrap-openjdk11-arm-12-11.0.5.10.1.tar.xz
Thanks Mikael. I've copied your bootstrap over into the same distribution directory where I've got the others. In a few hours you should be able to test the bootstrap change to make sure it works.
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=362eb92da22191ba1aa95524f86a62afc9d92496 commit 362eb92da22191ba1aa95524f86a62afc9d92496 Author: Mikael Urankar <mikael@FreeBSD.org> AuthorDate: 2021-05-18 15:57:36 +0000 Commit: Mikael Urankar <mikael@FreeBSD.org> CommitDate: 2021-05-18 15:57:36 +0000 java/openjdk11: Add armv6 / armv7 support PR: 255662 Approved by: portmgr (tier 2 blanket) java/bootstrap-openjdk11/Makefile | 8 ++++---- java/bootstrap-openjdk11/distinfo | 2 ++ java/openjdk11/Makefile | 10 +++++----- 3 files changed, 11 insertions(+), 9 deletions(-)
I am trying this on armv6, and the bootstrap vm says that it is for armv7 only. Therefore, the build of openjdk11 on armv6 fails. -- Martin
I don't have access to an armv6 board right now, do you have a build log? The bootstrap seems correct: readelf -A bootstrap-openjdk11/bin/java | grep CPU_arch Tag_CPU_arch: ARM v6 file bootstrap-openjdk11/bin/java: ELF 32-bit LSB executable, ARM, EABI5 version 1 (FreeBSD), dynamically linked, interpreter /libexec/ld-elf.so.1, for FreeBSD 12.2 Thanks.
This first log is from an emulation using qemu-user-static. The failure at the "cd" command at the end is strange - that directory is perfectly reachable and writable. This is interesting (all with qemu-user-static): [0]# /usr/local/bootstrap-openjdk11/bin/javac -version Error occurred during initialization of VM Server VM is only supported on ARMv7+ VFP [1]# I'll try with a real Raspberry-Pi B+ a little later. -- Martin
Created attachment 227974 [details] log file when running portmaster java/openjdk11 with qemu-arm-static
Created attachment 227980 [details] RPI-B panic message when trying to install openjdk11 So the rpi-b simply panicked...
And furthermore on the RPI-B: [0]# /usr/local/bootstrap-openjdk11/bin/javac -version Error occurred during initialization of VM Server VM is only supported on ARMv7+ VFP [1]# -- Martin
I'm afraid that armv6 is not supported in server mode: https://github.com/battleblow/openjdk-jdk11u/blob/bsd-port/src/hotspot/cpu/arm/vm_version_arm_32.cpp#L238 I've built this bootstrap on an armv7 board inside an armv6 chroot and never triggered this issue.... It's possible to use the zero mode but it'll require a new bootstrap.
FreeBSD armv6 support is going to be retired before FreeBSD 15.0.
At least the build on armv7 gets pretty far nowadays. https://pkg-status.freebsd.org/ampere2/data/main-armv7-default/p0047206c1c4f_s666303f59/logs/openjdk11-11.0.24+8.1_1.log Does anybody have a clue what this is about? /wrkdirs/usr/ports/java/openjdk11/work/jdk11u-jdk-11.0.24-8-1/src/hotspot/cpu/arm/assembler_arm_64.cpp:47:24: error: no member named 'LogicalImmediate' in 'Assembler' 47 | inline bool Assembler::LogicalImmediate::has_equal_subpatterns(uintx imm, int size) { | ~~~~~~~~~~~^ Googling these terms only pointed me to the pkg fallout emails. :-)
(In reply to Ronald Klop from comment #11) This conditional doesn't work: https://github.com/battleblow/jdk11u/blob/bsd-port/make/hotspot/lib/CompileJvm.gmk#L111 "call isTargetCpu, bsd-arm" should probably be "call isTargetCpu, arm"
Created attachment 254105 [details] v1 This patch fixes the build on armv7
(In reply to Mikael Urankar from comment #13) Great patch. Thanks. Was busy on this also. Will test your patch this soon. Are all the changes in the patch like _MACHINE_CPUFUNC_H_ and __bool_true_false_are_defined also acceptable on non-armv7 architectures (amd64, aarch64, etc.) or do we need to make them conditional?