Closed kloczek closed 3 years ago
Hi, can you please post the contents of tests/core_stacktrace.log
? Also, what OS/arch are you using?
Linux x86_64. Here is the log file content:
# random seed: R02S2455cf171ef75f3d9ed977bc4f8e4c66
1..3
# Start of stacktrace tests
# Start of core tests
ok 1 /stacktrace/core/to-json
ok 2 /stacktrace/core/from-json
lt-core_stacktrace: WEXITSTATUS: 1
FAIL core_stacktrace (exit status: 1)
Which is exctly the same what was in whole test suite output (because I've started tit with V=1 VERBOSE=1
)
Still can't reproduce, sorry. Could you please post the exact procedure how you got the sources, how you built them and how you made the tests fail?
E.g. when I download and extract https://github.com/abrt/satyr/releases/download/0.36/satyr-0.36.tar.gz and then run:
$ cd satyr-0.36
$ ./autogen.sh && ./configure && make && make check
Everything works fine. I'm testing on Fedora 33. Are you on the same OS?
I'm testing on Fedora 33. Are you on the same OS?
I'm using my own distribution which is mostly fedora rawhide based.
So you have no idea why that lt-core_stacktrace
testy unit finished with WEXITSTATUS: 1
?
I can reproduce this on Rawhide. It appears the dump_core
program from our test suite fails to run correctly.
So I'm trying on Rawhide running in podman:
make check
fails on the core_stacktrace
test.
tests/.libs/lt-core_stacktrace
segfaults when run by hand.
So far so good. But I can't debug it because running the binary in gdb
fails with "Error disabling address space randomization".
Turns out I have to rerun podman
with --cap-add=SYS_PTRACE --security-opt seccomp=unconfined
.
But then make check
finishes successfully but the binary still segfaults when run by hand. EDIT: This is to be expected.
Will look into it more tomorrow.
@kloczek, is this still an issue? I just tried with in F34 on an actual machine (not a container) and it seems to be working fine. It also builds in COPR, for F34 and Rawhide, and all the tests pass. If you're using docker or podman, please try passing the two args above. If you're on an actual physical machine and still this is happening, try commenting out the call to open
at tests/dump_core.c:58
. You should see the output from the core_stacktrace tests when you run make check
again.
Yep looks like it is OK. Closing PS. However iteresting what it was (?)