Description of problem: When building libint on aarch64, the following linking error occurs at final link stage: [...] .libs/hrr1_build_gh.o: In function `hrr1_build_gh(double const*, double*, double*, double*, int)': /builddir/build/BUILD/libint-0e0ffa7887e74e6ab1fb07c89be55f776c733731/src/lib/libint/src/hrr1_build_gh.cc:11:(.text+0x4c88): relocation truncated to fit: R_AARCH64_CALL26 against symbol `_hrr1_build_gh_0(double const*, double*, double*, double*, int)' defined in .text section in .libs/hrr1_build_gh.o /builddir/build/BUILD/libint-0e0ffa7887e74e6ab1fb07c89be55f776c733731/src/lib/libint/src/hrr1_build_gh.cc:12:(.text+0x4ca0): relocation truncated to fit: R_AARCH64_CALL26 against symbol `_hrr1_build_gh_1(double const*, double*, double*, double*, int)' defined in .text section in .libs/hrr1_build_gh.o Version-Release number of selected component (if applicable): 1.1.6-6.fc23 How reproducible: Always. Steps to Reproduce: 1. arm-koji build rawhide git://pkgs.fedoraproject.org/libint?#119a604475eaa95f2cfcfd8e2fb4132c7e989003 Additional information: http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3176102
gcc bug?
Looks like linker bug: https://sourceware.org/bugzilla/show_bug.cgi?id=18668
Bug in binutils (2.25.1-4.fc24). Looks like it's fixed upstream already, so please backport.
This is blocking the aarch64 build of cp2k and other stuff that depends on libint.
Hi Dominik, Please try out binutils-2.25.1-6.fc24 Cheers Nick
libint-1.1.6-6.fc24 built successfully, thanks Nick! http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3177887
(In reply to Nick Clifton from comment #5) > Hi Dominik, > > Please try out binutils-2.25.1-6.fc24 Can we get this fix in F-23 too please?
Hi Peter, > Can we get this fix in F-23 too please? I think that Fedora 23 is frozen at the moment... Cheers Nick
(In reply to Nick Clifton from comment #8) > Hi Peter, > > > Can we get this fix in F-23 too please? > > I think that Fedora 23 is frozen at the moment... Updates can still be submitted and they go to f23-updatest-testing as per a standard cycle. They're held there until Beta gets signed off and which point they continue on their way post beta.
OK then, please await the arrival of: binutils-2.25-14.fc23 Cheers Nick
> binutils-2.25-14.fc23 Can this please be submitted as an update to F-23 please?
binutils-2.25-14.fc23 has been submitted as an update to Fedora 23. https://bodhi.fedoraproject.org/updates/FEDORA-2015-3ed7d1146a
binutils-2.25-14.fc23 has been pushed to the Fedora 23 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with $ su -c 'dnf --enablerepo=updates-testing update binutils' You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2015-3ed7d1146a
Patch fails to apply: + echo 'Patch #17 (binutils-2.25.1-dynamic_list.patch):' Patch #17 (binutils-2.25.1-dynamic_list.patch): + /usr/bin/patch -p1 -b --suffix .dynamic_list~ --fuzz=0 patching file gold/layout.cc patching file gold/options.cc patching file gold/symtab.h patching file gold/testsuite/Makefile.am patching file gold/testsuite/Makefile.in Hunk #1 succeeded at 5266 (offset -11 lines). + echo 'Patch #18 (binutils-2.25-aarch64-pr18668.patch):' Patch #18 (binutils-2.25-aarch64-pr18668.patch): + /usr/bin/patch -p1 --fuzz=0 patching file bfd/elfnn-aarch64.c Hunk #2 FAILED at 4394. 1 out of 2 hunks FAILED -- saving rejects to file bfd/elfnn-aarch64.c.rej
Created attachment 1078594 [details] fix for binutils 2.25-14 failure
Hi Guys, This is very strange - I cannot reproduce this failure. In fact applying Marcin's patch actually breaks the build for me. Peter - does Marcin's patch work for you ? If so then there must be something wrong with my build system, although I have no idea what. Cheers Nick
Nick: f24 is fine as it is, f23 fails without that patch.
(In reply to Marcin Juszkiewicz from comment #17) > Nick: f24 is fine as it is, f23 fails without that patch. If you say so - but I have tried several times using the F23 sources and they always work for me. Cheers Nick
http://arm.koji.fedoraproject.org/koji/taskinfo?taskID=3208788 failed on applying: Patch #18 (binutils-2.25-aarch64-pr18668.patch): + /usr/bin/patch -p1 --fuzz=0 patching file bfd/elfnn-aarch64.c Hunk #2 FAILED at 4394. 1 out of 2 hunks FAILED -- saving rejects to file bfd/elfnn-aarch64.c.rej error: Bad exit status from /var/tmp/rpm-tmp.Bltrvp (%prep)
Ah - found it - the problem only occurs with an aarch64 native or cross build. If building for any other target a previous patch is not applied (patch16) and this meant that patch18 did apply. I have now checked in a fix to always apply patch16, even if the target is not aarch64. (The patch only affects aarch64 files, so this should not be a problem). Then I updated patch18 with Marcin's fix, so that now the F23 binutils should build for all targets. This fix is in: binutils-2.25-15.fc23 Cheers Nick
binutils-2.25-15.fc23 has been submitted as an update to Fedora 23. https://bodhi.fedoraproject.org/updates/FEDORA-2015-34e344e965
binutils-2.25-15.fc23 has been pushed to the Fedora 23 testing repository. If problems still persist, please make note of it in this bug report. If you want to test the update, you can install it with $ su -c 'dnf --enablerepo=updates-testing update binutils' You can provide feedback for this update here: https://bodhi.fedoraproject.org/updates/FEDORA-2015-34e344e965
binutils-2.25-15.fc23 has been pushed to the Fedora 23 stable repository. If problems still persist, please make note of it in this bug report.