Comment 4 for bug 2051925

Revision history for this message
Paul Mars (upils) wrote :

I am struggling to tell the builder to run the dh_auto_test as root.

I tried "export DEB_GAIN_ROOT_CMD='sudo -nE --'" in d/rules. And I tried to make sense of the documentation regarding Rules-Requires-Root but this is very confusing and so far I failed.

Aside from this, I added 2 autopkgtest rules to run the test suite and the memory tests. Both currently fail.

Memory mapped buffers not supported

Suite: tracefs library
  Test: Test tracefs/debugfs mounting ...passed
  Test: trace cpu read ...passed
  Test: trace cpu read_buf_percent ...FAILED
    1. tracefs-utest.c:1359 - data->done == true
    2. tracefs-utest.c:1301 - kbuf != NULL
    3. tracefs-utest.c:1359 - data->done == true
    4. tracefs-utest.c:1301 - kbuf != NULL
    5. tracefs-utest.c:1359 - data->done == true
    6. tracefs-utest.c:1301 - kbuf != NULL
    7. tracefs-utest.c:1359 - data->done == true
    8. tracefs-utest.c:1301 - kbuf != NULL
  Test: trace cpu pipe ...passed
  Test: trace pid events filter ...passed
  Test: trace pid function filter ...passed
  Test: trace sql ...passed
  Test: trace sql trace onmax ...passed
  Test: trace sql trace onchange ...passed
  Test: trace sql snapshot onmax ...passed
  Test: trace sql snapshot onchange ...passed
  Test: trace sql save onmax ...passed
  Test: trace sql save onchange ...passed
  Test: trace sql trace and snapshot onmax ...passed
  Test: trace sql trace and snapshot onchange ...passed
  Test: tracing file / directory APIs ...passed
  Test: instance file / directory APIs ...passed
  Test: instance file descriptor ...passed
  Test: instance reset ...passed
  Test: systems and events APIs ...passed
  Test: tracefs_iterate_snapshot_events API ...passed
  Test: tracefs_iterate_raw_events API ...passed
  Test: Follow events ...passed
  Test: Follow events clear ...passed
  Test: tracefs_tracers API ...passed
  Test: tracefs_local events API ...passed
  Test: tracefs_instances_walk API ...passed
  Test: tracefs_get_clock API ...passed
  Test: tracing on / off ...passed
  Test: tracing options ...passed
  Test: custom system directory ...passed
  Test: ftrace marker ...passed
  Test: kprobes ...passed
  Test: synthetic events ...passed
  Test: eprobes ...passed
  Test: uprobes ...FAILED
    1. tracefs-utest.c:2205 - ret == 0
    2. tracefs-utest.c:2205 - ret == 0

Run Summary: Type Total Ran Passed Failed Inactive
              suites 1 1 n/a 0 0
               tests 36 36 34 2 0
             asserts 22851466 22851466 22851456 10 n/a

Elapsed time = 24.581 seconds

Even tough actually running tests is definitely an improvement, I do not know if we want to MIR a package with already failing tests. I will try to investigate but is this considered blocking for the MIR process?