Closed pemensik closed 3 years ago
Every single test run prints:
*** Please report this problem to kyua-discuss@googlegroups.com detailing what you were doing before the crash happened; if possible, include the log file mentioned above
munmap_chunk(): invalid pointer
*** Fatal signal 6 received
Can you reproduce this in valgrind?
Sure, 100% reliable. What options would you like?
Okay, maybe I was wrong. I might be fixed on rawhide already, new kyua rebuild seems to pass tests on all architectures again. Not sure which part was fixed, but it passes on test machine with gcc-c++-11.0.0-0.19.fc35.ppc64le
.
It still fails on stacktrace test, but passes fine otherwise on all platforms:
stacktrace_test:dump_stacktrace__cannot_find_core -> passed [0.410s]
stacktrace_test:dump_stacktrace__cannot_find_gdb -> passed [0.828s]
stacktrace_test:dump_stacktrace__gdb_fail -> passed [0.879s]
stacktrace_test:dump_stacktrace__gdb_timeout -> passed [2.140s]
stacktrace_test:dump_stacktrace__integration -> failed: Line 391: atf::utils::grep_file("#0", exit_handle.stderr_file().str()) not met [0.649s]
stacktrace_test:dump_stacktrace__ok -> failed: Line 419: atf::utils::grep_file("^frame 1$", exit_handle.stderr_file().str()) not met [0.689s]
stacktrace_test:dump_stacktrace_if_available__append -> failed: Line 559: atf::utils::grep_file("frame 1", exit_handle.stderr_file().str()) not met [0.699s]
stacktrace_test:dump_stacktrace_if_available__no_coredump -> passed [0.011s]
stacktrace_test:dump_stacktrace_if_available__no_status -> passed [0.110s]
stacktrace_test:find_core__found__long -> failed: Core dumped, but no candidates found [0.729s]
stacktrace_test:find_core__found__short -> failed: Core dumped, but no candidates found [0.701s]
stacktrace_test:find_core__not_found -> passed [0.010s]
stacktrace_test:find_gdb__bogus_value -> passed [0.009s]
stacktrace_test:find_gdb__search_builtin__fail -> passed [0.002s]
stacktrace_test:find_gdb__search_builtin__ok -> passed [0.002s]
stacktrace_test:find_gdb__use_builtin -> passed [0.009s]
stacktrace_test:unlimit_core_size -> passed [0.730s]
stacktrace_test:unlimit_core_size__hard_is_zero -> failed: Line 281: !status.coredump() not met [0.359s]
I think default integration with systemd is responsible:
# cat /proc/sys/kernel/core_pattern
|/usr/lib/systemd/systemd-coredump %P %u %g %s %t %c %h
Could failing tests be skipped if core_pattern starts with '|'?
@jmmv, could you please rebuild the package on Fedora? I have filled bug #1926779, but no response from any kyua maintainers were received.
Fedora Rawhide recently upgraded to GCC 11 and rebuilt kyua with it. Because stacktrace_test is always failing on Fedora, unit tests were disabled. That is unfortunate, because kyua became unusable on rawhide. It fails on platforms: armv7hl i686 ppc64le s390x, ppc64le only few test fail. Others fail completely.
Check last rawhide build build.log for alternative architectures. Every single unit test failed, but it was silently ignored. If you check build log
My backtraces in bug report hint, it might be related to compile-time warning:
Backtrace from the bug:
Was kyua tested on any different distribution with the recent GCC? Do you have any ideas, how to fix this issue? I use kyua to ensure ISC BIND package is good. Unfortunately only my bind build. failed hard and revealed the problem. But would have to turn off unit tests, unless kyua is fixed soon. Any ideas, how to fix this?