* plans/ci: setting "how: beakerlib" in execute step is deprecated.
changed to "how: tmt" to comply with newer tmt versions.
tmt versions.
* ltrace with both --o and --c does not produce output to file:
add malloc to the list of syscalls to check for tracing.
* ltrace crashes when run on certain processes with --S option:
make pgrep to exact match to avoid getting two PIDs, one from
dbus-broker and other from dbus-broker-launch.
* ltrace doesn't work on PIE-binaries:
Added missing 32-bit dependencies.
Usually, this test is used for gating ltrace. In such case the
Fedora infrastructure offers the SRPM along with other ltrace
subrpms within a repo called "test-artifacts".
However, this test is also used to test dejagnu, because ltrace
has dejagnu based testsuite. In such case the "test-artifacts"
repo has dejagnu, but not ltrace.
We need to be able to get the ltrace SRPM in both cases.
"koji download-build" is a wrong tool to download artifacts being
tested. They are freshly scratch-built and hosted in a local repo
called "test-artifacts" unknown to koji. Using "dnf download"
directly and limited to that specific repo is a correct way to
reach them.
I'm adding a fallback using the repo's local files just in case
"dnf" suffers some temporary hiccup.
Note: In RHEL, it's "testsuite2" outside of categories, here I'm adding
it to the Sanity category.
Note: I've filed https://bugzilla.redhat.com/show_bug.cgi?id=1963110
and included it as a known failure to make this test pass.
To make the bz1427140-ltrace-ltrace-0-7-91-unwind-elfutils-patch work
not only for regular builds but for scratch builds as well, we need
to know how to refer them. This commit is just a temporary attempt to
find the environment variable holding the reference. I'll make use of
it in a follow-up commit.
If the kernel supports it AND glibc supports it, the ppc64le
syscalls may use the SCV variant instead of the SC variant.
The registers we care about have the same use so little change
is required.
Reference: RHEL BZ 1922305
GCC cannot follow the logic that determines if it's safe to free
lrhs in expr.c. Initializing these variables to NULL silences
this warning.
Resolves: #1799619
Tests on ppc64le are hanging koji, despite the forced timeout.
The rpmbuild finishes normally - even building rpms - but the
hung testcase keeps koji from seeing it as "done".
So skip tests on ppc64le for now, until we can fix the hang.
- Add fixes in handling of bogus paths that come from XDG_CONFIG_DIRS.
- Testsuite fixes for pre-installed config files.
- Extra AARCH64 includes for gcc 9.
- Testsuite: AARCH64 ifuncs not supported yet yet.
