[LTP] Did it pass or did it fail

[Cyril Hrubis]

Hi!
> I have a question about BROK and LTP results.  When running LTP I see
> this in the log file under the results directory:
> 
> madvise09                      FAIL       2
> 
> Which would lead one to believe that the madvise09 test failed.
> 
> 
> 
> But in the output from runltp I see this:
> 
> <<<test_start>>>
> tag=madvise09 stime=1490213376
> cmdline="madvise09"
> contacts=""
> analysis=exit
> <<<test_output>>>
> tst_test.c:847: INFO: Timeout per run is 0h 05m 00s
> madvise09.c:181: PASS: MADV_FREE pages were not freed immediatelly
> madvise09.c:187: INFO: Setting memory limits to 8388608 16777216
> safe_file_ops.c:144: BROK: Failed to open FILE '/sys/fs/cgroup/memory/ltp_madvise09_20474//memory.memsw.limit_in_bytes' for reading at madvise09.c:190: ENOENT
> madvise09.c:289: BROK: Child exitted unexpectedly

This means that the test had an internal error. Looking at the message
it looks like memory cgroup file is missing. What kernel version do you
have?

> Summary:
> passed   1
> failed   0
> skipped  0
> warnings 0
> <<<execution_status>>>
> initiation_status="ok"
> duration=0 termination_type=exited termination_id=2 corefile=no
> cutime=0 cstime=0
> <<<test_end>>>
> 
> 
> This is less clear.  There is the passed line on the one hand and the
> 'failed 0' line, but then there is the termination_id on the other hand and
> the BROK output as well.  Did the test pass or fail?  Is output of this type
> expected, where the log shows a FAIL but the stdout from runltp doesn't
> necessarily show a clear TFAIL or TPASS in the output?

I agree that the output is a bit confusing here. The TBROK means that
something happened in the test initialization and the test execution has
to be stopped, hence the test library cleans up and exits with TBROK (2)
exit status, which is what you see under the termination_id. Now the
ltp-pan, that executes the testcases, maps anything that exits with
non-zero (and is not TCONF (32)) to a failure. Hence you see that the
test has failed with exit status 2.

There is a work being done on a replacement for the runltp script and
ltp-pan and proof of concept is basically done, so we will get rid of
these reasonably soon anyway...

-- 
Cyril Hrubis
chrubis@suse.cz