Created attachment 235978 [details] Update to devel/binutils to 2.39 Hi portmgr@, Please perform an exp-run on the attached update to devel/binutils 2.39.
(In reply to Cy Schubert from comment #0) On my own desktop set only multimedia/libxine fails to link with new binutils /usr/local/bin/ld.bfd: xine-list.o: non-canonical reference to canonical protected function `xine_set_flags' in ../src/xine-engine/.libs/libxine.so /usr/local/bin/ld.bfd: failed to set dynamic section sizes: bad value cc: error: linker command failed with exit code 1 (use -v to see invocation) https://build.dimapanov.com/data/131amd64-dimaports/2022-08-18_14h58m41s/logs/errors/libxine-1.2.12_2.log
There was already an exp-run in PR #261875 with results posted. Could the failures be handled before doing another one? (binutils exp-runs are more intensive than some others)
(In reply to Antoine Brodin from comment #2) That was for 2.38. This PR is for 2.39.
(In reply to Antoine Brodin from comment #2) That was for 2.38. This PR is for 2.39. Also, the URLs to the package builds don't work.
Info pages have to be put in the INFO macro, not in the plist.
Created attachment 236016 [details] Fixes a INFO plist issue and addresses breakage in PR/261875 https://www.mail-archive.com/u-boot@lists.denx.de/msg430287.html discusses that GCC11 or greater is required but if <= GCC10 is used binutils must be built with --with-isa-spec=2.2. This patch adds that CONFIGURE_ARGS option. If this causes other problems we will need to rethink our minimum GCC requirement on those architectures that need GCC or conditionally add this option if GCC < 11 (though I don't see an easy way to test for that in ports other than gcc -v 2>&1 | awk '$0 ~ /gcc version/ {print $3}', which is ugly).
2.39 addresses the regressions in 2.38. Can we run an exp-run, please?
New failures logs on 12.3 i386: http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/amd64-binutils-2.39,1.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/elf-dissector-0.0.1_1.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/libxine-1.2.12_2.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/atf-rk3328-v2.6.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/atf-rk3399-v2.6.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/atf-sun50i_a64-v2.6.log http://pb2.nyi.freebsd.org/data/123i386-default-foo/2022-08-30_07h32m52s/logs/errors/atf-sun50i_h6-v2.6.log Around 20 ports were skipped due to those new failures
New failures logs on 12.3 amd64: http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/amd64-binutils-2.39,1.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/elf-dissector-0.0.1_1.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/libxine-1.2.12_2.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/atf-fvp-v2.6.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/atf-rk3328-v2.6.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/atf-rk3399-v2.6.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/atf-sun50i_a64-v2.6.log http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/atf-sun50i_h6-v2.6.log
New failures logs on 13.0 i386: http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/amd64-binutils-2.39,1.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/elf-dissector-0.0.1_1.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/libxine-1.2.12_2.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/atf-rk3328-v2.6.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/atf-rk3399-v2.6.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/atf-sun50i_a64-v2.6.log http://package23.nyi.freebsd.org/data/130i386-default-foo/2022-08-26_08h04m06s/logs/errors/atf-sun50i_h6-v2.6.log
1. The plist breakage is not reproducable with this version. 2. I am not able to reproduce the build breakage in 13-STABLE (amd64/i386) jail. 3. No such breakage under 12-STABLE (amd64/i386). I will attach a new diff based on the current state of the ports tree.
(In reply to Cy Schubert from comment #11) It was reproduced on all the exp builders.
Having built binutils 2.39 under poudriere 14-CURRENT, 13.1-RELEASE, 12.3-RELEASE, in i386 and amd64 poudriere jails, using a ports tree as of this morning I have seen no failures. I'm puzzled why the package builder poudriere is experiencing these failures. The only reason we see package failures is that only a subset of files were successfully built. Can I have a copy of each failed wrkdir. ./config.log might shed some light on why the package build poudriere jobs all failed while all mine succeeded.
(In reply to Cy Schubert from comment #13) No, it's because the plist of all the binutils flavors was not checked, binutils has around 20 flavors, it's not just native.
The build problems appear to be FLAVORs under i386. Probably due to 32-bit issues in which some files are not built and therefore not installed. I've seen this behavour before with other ports. Fix the build problems and the pkg-plist problems magically disappear. Building natively is not a problem however.
(In reply to Cy Schubert from comment #15) this is a failure log for amd64 flavor on amd64: http://package18.nyi.freebsd.org/data/123amd64-default-foo/2022-08-26_08h04m01s/logs/errors/amd64-binutils-2.39,1.log
(In reply to Antoine Brodin from comment #16) This is certainly not consistent with my amd64 flavor build in a 12.3 amd64 poudriere jail. (Sorry, my poudriere server is not internet accessible.) =>> Building devel/binutils build started at Sun Oct 9 12:58:04 PDT 2022 port directory: /usr/ports/devel/binutils package name: binutils-2.39,1 building for: FreeBSD kq386 12.3-RELEASE-p7 FreeBSD 12.3-RELEASE-p7 amd64 maintained by: cy@FreeBSD.org Makefile ident: Poudriere version: 3.3.7_1 Host OSVERSION: 1400072 Jail OSVERSION: 1203000 Job Id: 01 ---Begin Environment--- SHELL=/bin/csh OSVERSION=1203000 UNAME_v=FreeBSD 12.3-RELEASE-p7 UNAME_r=12.3-RELEASE-p7 BLOCKSIZE=K MAIL=/var/mail/root POUDRIERE_ETC=/var/db/local/poudriere-bob STATUS=1 HOME=/root PATH=/sbin:/bin:/usr/sbin:/usr/bin:/usr/local/sbin:/usr/local/bin:/root/bin LOCALBASE=/usr/local USER=root LIBEXECPREFIX=/usr/local/libexec/poudriere POUDRIERE_VERSION=3.3.7_1 MASTERMNT=/poudriere/bob/data/.m/12R3amd64-cy-git/ref POUDRIERE_BUILD_TYPE=bulk PACKAGE_BUILDING=yes SAVED_TERM=screen PWD=/poudriere/bob/data/.m/12R3amd64-cy-git/ref/.p/pool P_PORTS_FEATURES=FLAVORS SELECTED_OPTIONS MASTERNAME=12R3amd64-cy-git SCRIPTPREFIX=/usr/local/share/poudriere OLDPWD=/poudriere/bob/data/.m/12R3amd64-cy-git/ref/.p SCRIPTPATH=/usr/local/share/poudriere/bulk.sh POUDRIEREPATH=/usr/local/bin/poudriere ---End Environment--- ---Begin Poudriere Port Flags/Env--- PORT_FLAGS= PKGENV= FLAVOR=native DEPENDS_ARGS= MAKE_ARGS= FLAVOR=native ---End Poudriere Port Flags/Env--- ---Begin OPTIONS List--- ===> The following configuration options are available for binutils-2.39,1: NLS=on: Native Language Support RELRO=off: Enable relocation memory corruption mitigation (-z relro) by default STATIC=off: Build static executables and/or libraries ===> Use 'make config' to modify these settings ---End OPTIONS List--- [...] fi gmake[4]: Entering directory '/wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39/libiberty/testsuite' gmake[4]: Nothing to be done for 'install'. gmake[4]: Leaving directory '/wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39/libiberty/testsuite' gmake[3]: Leaving directory '/wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39/libiberty' gmake[2]: Nothing to be done for 'install-target'. gmake[2]: Leaving directory '/wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39' gmake[1]: Leaving directory '/wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39' install -m 0644 /wrkdirs/usr/ports/devel/binutils/work-native/binutils-2.39/include/plugin-api.h /wrkdirs/usr/ports/devel/binutils/work-native/stage/usr/local/include/ /bin/rm -f /wrkdirs/usr/ports/devel/binutils/work-native/stage/usr/local/include/ansidecl.h /usr/bin/sed -i.bak '/#include "ansidecl.h"/d' /wrkdirs/usr/ports/devel/binutils/work-native/stage/usr/local/include/bfd.h ====> Compressing man pages (compress-man) =========================================================================== =======================<phase: package >============================ ===> Building package for binutils-2.39,1 =========================================================================== =>> Cleaning up wrkdir ===> Cleaning for binutils-2.39,1 build of devel/binutils | binutils-2.39,1 ended at Sun Oct 9 13:09:29 PDT 2022 build time: 00:11:26 The only difference is mine says, MAKE_ARGS= FLAVOR=native while the failed build has MAKE_ARGS= FLAVOR=amd64. amd64 using FLAVOR=native works while amd64 with FLAVOR=amd64 does not. I'll try that out. Though FLAVOR=native and FLAVOR=amd64 building on an amd64 should produce the same result. But maybe not. Interesting.
(In reply to Cy Schubert from comment #17) The log above has FLAVOR=native , not FLAVOR=amd64
I've now been able to duplicate this from the command line. make check-orphans completes successfully on amd64. make FLAVOR=amd64 check-orphans fails on amd64. Having been able to duplicate the problem from the command line solving this interactively will be much easier now.
Created attachment 237383 [details] Can you try this patch please? Can you try this patch please. plist regressions have been fixed.
Keyword: patch or patch-ready – in lieu of summary line prefix: [patch] * bulk change for the keyword * summary lines may be edited manually (not in bulk). Keyword descriptions and search interface: <https://bugs.freebsd.org/bugzilla/describekeywords.cgi>
What about the elf-dissector / libxine / atf-* failures? Were they fixed? (I see 0 commit mentioning this bug report)
The atf-* ports now build correctly. I'll post patches later. elf-dissector may need the attention of our upstream because INIT_DISASSEMBLE_INFO() now has a fourth argument. libxine will likely need the patch from https://www.linuxquestions.org/questions/slackware-14/regression-on-current-with-xine-lib-4175715727/. This hasn't been tested yet.
The libxine patch works as well. Will post later. This leaves elf-dissector with the extra argument to INIT_DISASSEMBLE_INFO().
Created attachment 237731 [details] This patch should address reported regressions. Please try this patch.
Exp-run looks fine
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=1d2e6d5c44d0580f53a830ad74b85ec444952b40 commit 1d2e6d5c44d0580f53a830ad74b85ec444952b40 Author: Cy Schubert <cy@FreeBSD.org> AuthorDate: 2022-10-25 03:17:11 +0000 Commit: Cy Schubert <cy@FreeBSD.org> CommitDate: 2022-11-07 18:38:55 +0000 sysutils/atf-*: Fix binutils 3.39 build PR: 265917 sysutils/atf-master/Makefile | 2 ++ 1 file changed, 2 insertions(+)
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=4b9324b680ae0adabf23d962ba624f66ded51b81 commit 4b9324b680ae0adabf23d962ba624f66ded51b81 Author: Cy Schubert <cy@FreeBSD.org> AuthorDate: 2022-10-25 14:16:02 +0000 Commit: Cy Schubert <cy@FreeBSD.org> CommitDate: 2022-11-07 18:38:56 +0000 devel/elf-dissector: Fix binutils 2.39 regression Fix build with binutils 2.39 PR: 265917 .../files/patch-src_lib_disassembler_disassembler.cpp (new) | 11 +++++++++++ 1 file changed, 11 insertions(+)
A commit in branch main references this bug: URL: https://cgit.FreeBSD.org/ports/commit/?id=eab31cc222286b7ba79a629da7200b0bffe63f03 commit eab31cc222286b7ba79a629da7200b0bffe63f03 Author: Cy Schubert <cy@FreeBSD.org> AuthorDate: 2022-10-25 04:12:09 +0000 Commit: Cy Schubert <cy@FreeBSD.org> CommitDate: 2022-11-07 18:38:55 +0000 multimedia/libxine: Fix binutils 2.39 regression Fix build under 2.39. PR: 265917 Obtained from: https://www.linuxquestions.org/questions/slackware-14/\ regression-on-current-with-xine-lib-4175715727/ multimedia/libxine/Makefile | 2 ++ 1 file changed, 2 insertions(+)
Good stuff, thanks.
devel/binutils is update to 2.39 with ports 32d885a48f08.