Re: [PATCH v10 bpf-next] bpf/selftests: fold test_current_pid_tgid_new_ns into test_progs.

On Mon, Dec 21, 2020 at 2:25 PM Carlos Neira [off-list ref] wrote:

Currently tests for bpf_get_ns_current_pid_tgid() are outside test_progs.
This change folds test cases into test_progs.

Changes from v9:

 - Added test in root namespace.
 - Fixed changed tracepoint from sys_enter to sys_usleep.
 - Fixed pid, tgid values were inverted.
 - Used CLONE(2) for namespaced test, the new process pid will be 1.
 - Used ASSERTEQ on pid/tgid validation.
 - Added comment on CLONE(2) call

Signed-off-by: Carlos Neira <redacted>
---

See checkpatch.pl errors ([0]), ignore the "do not initialize globals
with zero" ones. Next time, please don't wait for me to point out
every single instance where you didn't align wrapped around
parameters.

  [0] https://patchwork.hopto.org/static/nipa/405025/11985347/checkpatch/stdout

 tools/testing/selftests/bpf/.gitignore        |   1 -
 tools/testing/selftests/bpf/Makefile          |   3 +-
 .../bpf/prog_tests/ns_current_pid_tgid.c      | 149 ++++++++++------
 .../bpf/progs/test_ns_current_pid_tgid.c      |  29 ++--
 .../bpf/test_current_pid_tgid_new_ns.c        | 160 ------------------
 5 files changed, 106 insertions(+), 236 deletions(-)
 delete mode 100644 tools/testing/selftests/bpf/test_current_pid_tgid_new_ns.c

[...]

-       id = (__u64) tid << 32 | pid;
-       bss.user_pid_tgid = id;
+cleanup:
+        test_ns_current_pid_tgid__destroy(skel);
+}
+
+static int newns_pidtgid(void *arg)

newns_pidtgid and test_ns_current_pid_tgid_global_ns look identical to
me, am I missing something on why you didn't reuse the testing logic
between the two?

+{
+       struct test_ns_current_pid_tgid__bss  *bss;
+       int err = 0, duration = 0;
+       struct test_ns_current_pid_tgid *skel;
+       pid_t pid, tgid;
+       struct stat st;

[...]