| Summary: | kde-apps/kate-lib-24.08.3 tests segfault | ||
|---|---|---|---|
| Product: | Gentoo Linux | Reporter: | Paolo Pedroni <paolo.pedroni> |
| Component: | Current packages | Assignee: | Gentoo KDE team <kde> |
| Status: | RESOLVED FIXED | ||
| Severity: | normal | CC: | kostadinshishmanov, paolo.pedroni |
| Priority: | Normal | Keywords: | PATCH, TESTFAILURE |
| Version: | unspecified | ||
| Hardware: | All | ||
| OS: | Linux | ||
| Whiteboard: | |||
| Package list: | Runtime testing required: | --- | |
| Bug Depends on: | |||
| Bug Blocks: | 914580 | ||
| Attachments: | kate-lib-24.08.3:20241125-123807.log.gz | ||
|
Description
Paolo Pedroni
2024-11-25 12:45:24 UTC
0x0000000000000018 is obviously suspicious. Can you try narrow down flags? (In reply to Sam James from comment #1) > 0x0000000000000018 is obviously suspicious. Can you try narrow down flags? The only flag I could remove is -mshstk, and without that tests just take longer to fail: 5/8 Testing: kateapp-location_history_test 5/8 Test: kateapp-location_history_test Command: "/var/tmp/portage/kde-apps/kate-lib-24.08.3/work/kate-24.08.3_build/bin/location_history_test" "-platform" "offscreen" Directory: /var/tmp/portage/kde-apps/kate-lib-24.08.3/work/kate-24.08.3_build/apps/lib/autotests "kateapp-location_history_test" start time: Nov 25 16:01 CET Output: ---------------------------------------------------------- kf.config.core: Watching absolute paths is not supported "/var/tmp/portage/kde-apps/kate-lib-24.08.3/temp/kate-fVtBjj/testconfigfilerc" <end of output> Test time = 168.90 sec ---------------------------------------------------------- Test Failed. "kateapp-location_history_test" end time: Nov 25 16:04 CET "kateapp-location_history_test" time elapsed: 00:02:48 ---------------------------------------------------------- 6/8 Testing: kateapp-kate_view_mgmt_tests 6/8 Test: kateapp-kate_view_mgmt_tests Command: "/var/tmp/portage/kde-apps/kate-lib-24.08.3/work/kate-24.08.3_build/bin/kate_view_mgmt_tests" "-platform" "offscreen" Directory: /var/tmp/portage/kde-apps/kate-lib-24.08.3/work/kate-24.08.3_build/apps/lib/autotests "kateapp-kate_view_mgmt_tests" start time: Nov 25 16:01 CET Output: ---------------------------------------------------------- kf.config.core: Watching absolute paths is not supported "/var/tmp/portage/kde-apps/kate-lib-24.08.3/temp/kate-OYVXIe/testconfigfilerc" ********* Start testing of KateViewManagementTests ********* Config: Using QtTest library 6.7.3, Qt 6.7.3 (x86_64-little_endian-lp64 shared (dynamic) release build; by Clang 18.1.8), gentoo 2.17 PASS : KateViewManagementTests::initTestCase() Received signal 11 (SIGSEGV), code 1, for address 0x0000000000000018 testSingleViewspaceDoesntCloseWhenLastViewClosed function time: 7ms, total time: 7ms === Stack trace === ptrace: Operation not permitted. === End of stack trace === <end of output> Test time = 168.90 sec ---------------------------------------------------------- Test Failed. "kateapp-kate_view_mgmt_tests" end time: Nov 25 16:04 CET "kateapp-kate_view_mgmt_tests" time elapsed: 00:02:48 ---------------------------------------------------------- It's worth noting that this started to happen after upgrading to gcc-14. Same thing with -march=x86-64-v4 and with -march=znver4 and no LTO. Excellent, thanks. Same version passed before? > *Config: Using QtTest library 6.7.3, Qt 6.7.3 (x86_64-little_endian-lp64 shared (dynamic) release build; by Clang 18.1.8), gentoo 2.17
Clang?
(In reply to Sam James from comment #3) > Excellent, thanks. Same version passed before? Yes, it did. Another possible difference is that the last time it worked, it was built against kde-frameworks-6.6.0, while now it is against kde-frameworks-6.7.0. > Clang? I use to build qt with clang, ever since we had qt6 in tree: never had any issue with it. But then it's not really obvious if it's a GCC problem or a Clang one. I can't really do much with this as-is. (In reply to Sam James from comment #6) > But then it's not really obvious if it's a GCC problem or a Clang one. I > can't really do much with this as-is. It used to work in the same conditions. Only difference is GCC and kde-frameworks version. I could revert to previous kde-frameworks and try again, but it's going to take a while. Do you think it may be worth to try? I understand, but I need to be able to reproduce it, and having a mix of things to try emulate locally isn't great. Especially as Qt has often had UB in the past, so it being built with Clang could be relevant. It would be more useful, if possible, to build KF and Qt with the same compiler and verify it happens, rather than downgrade. (In reply to Sam James from comment #8) > It would be more useful, if possible, to build KF and Qt with the same > compiler and verify it happens, rather than downgrade. I will try tomorrow. FWIW, these also segfault for me when I build kde-apps/kate-lib with portage, but if I try to build it "locally" in my home directory, the whole test suite passes with no failures. (In reply to Kostadin Shishmanov from comment #10) > FWIW, these also segfault for me when I build kde-apps/kate-lib with > portage, but if I try to build it "locally" in my home directory, the whole > test suite passes with no failures. That would mean either: a) sandbox; b) some other environment variable usually. If you run it in a terminal after sourcing the environment file in a subshell, does it fail? We discussed it on IRC. ``` [...] Thread 1 (Thread 0x7f2441d9e9c0 (LWP 51) "kate_view_mgmt_"): #0 0x00007f244a11bab7 in wait4 () at /usr/lib64/libc.so.6 #1 0x00007f244ab21ee5 in generateStackTrace () at /usr/src/debug/dev-qt/qtbase-6.7.3-r2/qtbase-everywhere-src-6.7.3/src/testlib/qtestcase.cpp:416 #2 0x00007f244ab22260 in generateStackTrace () at /usr/src/debug/dev-qt/qtbase-6.7.3-r2/qtbase-everywhere-src-6.7.3/src/testlib/qtestcase.cpp:380 #3 (anonymous namespace)::FatalSignalHandler::actionHandler (signum=11, info=<optimized out>) at /usr/src/debug/dev-qt/qtbase-6.7.3-r2/qtbase-everywhere-src-6.7.3/src/testlib/qtestcase.cpp:2217 #4 0x00007f244a0539e0 in <signal handler called> () at /usr/lib64/libc.so.6 #5 0x00007f2447920e15 in QVariant::Private::typeInterface (this=0x0, this=<optimized out>) at /usr/include/qt6/QtCore/qvariant.h:132 #6 QVariant::Private::type (this=0x0, this=<optimized out>) at /usr/include/qt6/QtCore/qvariant.h:137 #7 qvariant_cast<QDBusVariant> (v=...) at /usr/include/qt6/QtCore/qvariant.h:755 #8 QVariant::value<QDBusVariant>() const & (this=0x0) at /usr/include/qt6/QtCore/qvariant.h:517 #9 KColorSchemeWatcherXDG::KColorSchemeWatcherXDG (this=<optimized out>, this=<optimized out>) at /usr/src/debug/kde-frameworks/kguiaddons-6.8.0/kguiaddons-6.8.0/src/colors/kcolorschemewatcher_xdg.cpp:32 #10 0x00007f244790f1a6 in std::make_unique<KColorSchemeWatcherXDG> () at /usr/lib/gcc/x86_64-pc-linux-gnu/15/include/g++-v15/bits/unique_ptr.h:1076 #11 KColorSchemeWatcherPrivate::KColorSchemeWatcherPrivate (this=0x557d5696b210) at /usr/src/debug/kde-frameworks/kguiaddons-6.8.0/kguiaddons-6.8.0/src/colors/kcolorschemewatcher.cpp:36 #12 KColorSchemeWatcher::KColorSchemeWatcher (this=this@entry=0x557d56954638, parent=parent@entry=0x0, this=<optimized out>, parent=<optimized out>) at /usr/src/debug/kde-frameworks/kguiaddons-6.8.0/kguiaddons-6.8.0/src/colors/kcolorschemewatcher.cpp:45 [...] ``` https://invent.kde.org/frameworks/kguiaddons/-/commit/91c23a57f39103201d305480eb24039942a376ad fixes it The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=959e68fde158b58b374e27c7f171831e96096366 commit 959e68fde158b58b374e27c7f171831e96096366 Author: Andreas Sturmlechner <asturm@gentoo.org> AuthorDate: 2024-11-27 17:57:24 +0000 Commit: Andreas Sturmlechner <asturm@gentoo.org> CommitDate: 2024-11-27 18:02:45 +0000 kde-frameworks/kguiaddons: Don't try to access QDBusMessage if not ... ... successful reply Upstream commit 91c23a57f39103201d305480eb24039942a376ad Closes: https://bugs.gentoo.org/944927 Signed-off-by: Andreas Sturmlechner <asturm@gentoo.org> ...kguiaddons-6.7.0-dont-access-QDBusMessage.patch | 42 +++++++++++++++++ .../kguiaddons/kguiaddons-6.7.0-r1.ebuild | 52 ++++++++++++++++++++++ .../kguiaddons/kguiaddons-6.8.0-r1.ebuild | 52 ++++++++++++++++++++++ 3 files changed, 146 insertions(+) |
