Created attachment 904954 [details] build.log (amd64, 1.12.1) ninja-1.12.1 builds fine with ld.bfd linker but not with mold-2.34.0: [...] [29/34] x86_64-pc-linux-gnu-g++ -MMD -MT build/util.o -MF build/util.o.d -g -Wall -Wextra -Wno-deprecated -Wno-missing-field-initializers -Wno-unused-parameter -fno-rtti -fno-exceptions -std=c++11 -fvisibility=hidden -pipe '-DNINJA_PYTHON="python"' -fdiagnostics-color -DUSE_PPOLL -DNINJA_HAVE_BROWSE -I. -O2 -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -c src/util.cc -o build/util.o [30/34] x86_64-pc-linux-gnu-g++ -MMD -MT build/graph.o -MF build/graph.o.d -g -Wall -Wextra -Wno-deprecated -Wno-missing-field-initializers -Wno-unused-parameter -fno-rtti -fno-exceptions -std=c++11 -fvisibility=hidden -pipe '-DNINJA_PYTHON="python"' -fdiagnostics-color -DUSE_PPOLL -DNINJA_HAVE_BROWSE -I. -O2 -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -c src/graph.cc -o build/graph.o [31/34] x86_64-pc-linux-gnu-g++ -MMD -MT build/build.o -MF build/build.o.d -g -Wall -Wextra -Wno-deprecated -Wno-missing-field-initializers -Wno-unused-parameter -fno-rtti -fno-exceptions -std=c++11 -fvisibility=hidden -pipe '-DNINJA_PYTHON="python"' -fdiagnostics-color -DUSE_PPOLL -DNINJA_HAVE_BROWSE -I. -O2 -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -c src/build.cc -o build/build.o [32/34] x86_64-pc-linux-gnu-g++ -MMD -MT build/ninja.o -MF build/ninja.o.d -g -Wall -Wextra -Wno-deprecated -Wno-missing-field-initializers -Wno-unused-parameter -fno-rtti -fno-exceptions -std=c++11 -fvisibility=hidden -pipe '-DNINJA_PYTHON="python"' -fdiagnostics-color -DUSE_PPOLL -DNINJA_HAVE_BROWSE -I. -O2 -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -c src/ninja.cc -o build/ninja.o [33/34] rm -f build/libninja.a && x86_64-pc-linux-gnu-ar crs build/libninja.a build/browse.o build/depfile_parser.o build/lexer.o build/build.o build/build_log.o build/clean.o build/clparser.o build/debug_flags.o build/deps_log.o build/disk_interface.o build/dyndep.o build/dyndep_parser.o build/edit_distance.o build/eval_env.o build/graph.o build/graphviz.o build/json.o build/line_printer.o build/manifest_parser.o build/metrics.o build/missing_deps.o build/parser.o build/state.o build/status.o build/string_piece_util.o build/util.o build/version.o build/subprocess-posix.o d()x86_64-pc-linux-gnu-g++ -Lbuild -O2 -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -fuse-ld=mold -Wl,-O1 -Wl,--as-needed -Wl,-z,pack-relative-relocs -o ninja build/ninja.o -lninja Traceback (most recent call last): File "/var/tmp/portage/dev-build/ninja-1.12.1/work/ninja-1.12.1/configure.py", line 781, in <module> subprocess.check_call(rebuild_args) File "/usr/lib/python3.12/subprocess.py", line 413, in check_call raise CalledProcessError(retcode, cmd) subprocess.CalledProcessError: Command '['./ninja', '-v']' died with <Signals.SIGSEGV: 11>. * ERROR: dev-build/ninja-1.12.1::gentoo failed (compile phase): * Failed to run command: python3.12 configure.py --with-python=python --bootstrap --verbose
Created attachment 904955 [details] emerge --info
Can you run ./ninja -v under gdb in the build directory and get a backtrace? gdb --args ./ninja -v r bt
# gdb --args ./ninja -v GNU gdb (Gentoo 14.2 vanilla) 14.2 Copyright (C) 2023 Free Software Foundation, Inc. License GPLv3+: GNU GPL version 3 or later <http://gnu.org/licenses/gpl.html> This is free software: you are free to change and redistribute it. There is NO WARRANTY, to the extent permitted by law. Type "show copying" and "show warranty" for details. This GDB was configured as "x86_64-pc-linux-gnu". Type "show configuration" for configuration details. For bug reporting instructions, please see: <https://bugs.gentoo.org/>. Find the GDB manual and other documentation resources online at: <http://www.gnu.org/software/gdb/documentation/>. For help, type "help". Type "apropos word" to search for commands related to "word"... Reading symbols from ./ninja... (gdb) r Starting program: /var/tmp/portage/dev-build/ninja-1.12.1/work/ninja-1.12.1/ninja -v [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib64/libthread_db.so.1". [Detaching after vfork from child process 43233] [1/1] x86_64-pc-linux-gnu-g++ -Lbuild -Os -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -fuse-ld=mold -Wl,-O1 -Wl,--as-needed -Wl,-z,pack-relative-relocs -o ninja build/ninja.o -lninja [Inferior 1 (process 43230) exited normally] (gdb) bt
Starting program: /var/tmp/portage/dev-build/ninja-1.12.1/work/ninja-1.12.1/ninja -v [Thread debugging using libthread_db enabled] Using host libthread_db library "/usr/lib64/libthread_db.so.1". [Detaching after vfork from child process 43233] [1/1] x86_64-pc-linux-gnu-g++ -Lbuild -Os -march=znver3 -pipe -D_FILE_OFFSET_BITS=64 -fuse-ld=mold -Wl,-O1 -Wl,--as-needed -Wl,-z,pack-relative-relocs -o ninja build/ninja.o -lninja [Inferior 1 (process 43230) exited normally] (gdb) bt No stack. (gdb)
It's not segfaulting with your manual runs. Maybe sandbox-related? Can you reproduce the issue with FEATURES="-sandbox -usersandbox"?
Yes it still fails with FEATURES="-sandbox -usersandbox", but the error is different: [...] [33/34] rm -f build/libninja.a && x86_64-pc-linux-gnu-ar crs build/libninja.a build/browse.o build/depfile_parser.o build/lexer.o build/build.o build/build_log.o build/clean.o build/clparser.o build/debug_flags.o build/deps_log.o build/disk_interface.o build/dyndep.o build/dyndep_parser.o build/edit_distance.o build/eval_env.o build/graph.o build/graphviz.o build/json.o build/line_printer.o build/manifest_parser.o build/metrics.o build/missing_deps.o build/parser.o build/state.o build/status.o build/string_piece_util.o build/util.o build/version.o build/subprocess-posix.o terminate called after throwing an instance of 'std::bad_alloc' what(): std::bad_alloc Traceback (most recent call last): File "/var/tmp/portage/dev-build/ninja-1.12.1/work/ninja-1.12.1/configure.py", line 781, in <module> subprocess.check_call(rebuild_args) File "/usr/lib/python3.12/subprocess.py", line 413, in check_call raise CalledProcessError(retcode, cmd) subprocess.CalledProcessError: Command '['./ninja', '-v']' died with <Signals.SIGABRT: 6>. * ERROR: dev-build/ninja-1.12.1::gentoo failed (compile phase):
Do you get a segfault and/or an abort if you call ./ninja -v in the build directory without gdb? Also, it might be related to having the exact state of the build directory. For example, it might not fail if most of the 34 build steps have already been completed by a previous ninja run. To troubleshoot that, you would need to modify configure.py so that it does not call "./ninja -v" itself after bootstrapping.
It could also be triggered by something Portage sets in the environment.
*** Bug 941876 has been marked as a duplicate of this bug. ***
Reopening as we have another dupe: bug 941876.
FYI Rui had found this to be a change in the kernel that breaks the user space. Commit 2a010c412853 ("fs: don't block i_writecount during exec"), which went into mainline in the 6.11 cycle, changed the behavior of writing to a running executable that mold relies on. It doesn't seem clear whether the kernel side will back off the change. Meanwhile Rui has a temporary patch for these running mold on kernel v6.11. https://github.com/rui314/mold/issues/1361#issuecomment-2439432674
(In reply to Jinghao Jia from comment #11) > FYI Rui had found this to be a change in the kernel that breaks the user > space. Commit 2a010c412853 ("fs: don't block i_writecount during exec"), > which went into mainline in the 6.11 cycle, changed the behavior of writing > to a running executable that mold relies on. Thanks for taking this upstream and for your efforts! I wouldn't have thought a kernel change would be the root cause of the issue...
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=f6da81d9f85e2223a469e043dc95fd6fc555409f commit f6da81d9f85e2223a469e043dc95fd6fc555409f Author: Sam James <sam@gentoo.org> AuthorDate: 2024-10-28 11:36:23 +0000 Commit: Sam James <sam@gentoo.org> CommitDate: 2024-10-28 11:38:18 +0000 sys-devel/mold: workaround linux-6.11 incompatibility Don't write to existing files, instead replace them. This fixes (or works around) an incompatibility with linux-6.11. Bug: https://github.com/rui314/mold/issues/1361 Bug: https://bugs.gentoo.org/938659 Closes: https://bugs.gentoo.org/941070 Thanks-to: Jinghao Jia <jinghao7@illinois.edu> Thanks-to: ernsteiswuerfel <erhard_f@mailbox.org> Signed-off-by: Sam James <sam@gentoo.org> sys-devel/mold/files/mold-2.34.1-linux-6.11.patch | 22 +++++ sys-devel/mold/mold-2.34.1-r1.ebuild | 107 ++++++++++++++++++++++ sys-devel/mold/mold-9999.ebuild | 4 + 3 files changed, 133 insertions(+)
*** Bug 942516 has been marked as a duplicate of this bug. ***
