On MIPS I get: FAIL: run-backtrace-native-core.sh ================================== /<<PKGBUILDDIR>>/tests/backtrace: dwfl_pid: Unwinding not supported for this architecture backtrace-child-core.9086: arch not supported rmdir: failed to remove 'test-9078': Directory not empty FAIL run-backtrace-native-core.sh (exit status: 1)
That is indeed weird. It should exit 77 on that message. But it seems there is something left in that test-9078 test subdir. That would indeed make it fail. What is in that dir? Note that there is no MIPS backend support, so failures are expected. It might be that you have local debian patches that provide some MIPS support which might explain the issue. It would be nice to have a MIPS arch expert submit these patches upstream to get official MIPS support for elfutils.
Yes, I've had this mips patch written in 2006. But I've been unable to contact the author of that. It has had various fixes since, but the current version also has other issues, see: https://buildd.debian.org/status/fetch.php?pkg=elfutils&arch=mips&ver=0.175-1&stamp=1542667301&raw=0 Anyway, the directory contains: -rw------- 1 kroeckx kroeckx 8790016 Nov 20 19:23 core.10371 -rw-r--r-- 1 kroeckx kroeckx 0 Nov 20 19:23 core.10371.bt -rw-r--r-- 1 kroeckx kroeckx 102 Nov 20 19:23 core.10371.err
Odd, the trap on exit really should remove those. But explicitly cleaning the test files seems to work: diff --git a/tests/backtrace-subr.sh b/tests/backtrace-subr.sh index ff42c6f..53c719d 100644 --- a/tests/backtrace-subr.sh +++ b/tests/backtrace-subr.sh @@ -85,6 +85,7 @@ check_unsupported() testname=$2 if grep -q ': Unwinding not supported for this architecture$' $err; then echo >&2 $testname: arch not supported + test_cleanup exit 77 fi }
I can confirm that the patch fixes the issue for me.
commit 628b4a93c6863b9982d817db6acaacbc4e116453 Author: Mark Wielaard <mark@klomp.org> Date: Wed Nov 21 20:07:14 2018 +0100 tests: Call test_cleanup in backtrace-subr.sh check_unsupported. We want to make sure all (temporary) test files are cleaned up even when we exit 77 to skip the testcase. https://sourceware.org/bugzilla/show_bug.cgi?id=23901 Tested-by: Kurt Roeckx <kurt@roeckx.be> Signed-off-by: Mark Wielaard <mark@klomp.org>