On Fri, Sep 08, 2023 at 12:24:10PM +0200, Michal Simek wrote:
On 9/7/23 20:14, Tom Rini wrote:
On Thu, Sep 07, 2023 at 05:30:03PM +0200, Michal Simek wrote:
Hi Simon and Tom,
I am preparing pull request and I see that CI loop is reporting issue with sandbox trace and I have no idea what's going wrong here.
This is the first problematic commit but have no clue what it has to do with trace.
https://source.denx.de/u-boot/custodians/u-boot-microblaze/-/commit/8c5f80d1...
If you have an access you can take a look at my queue to see that only this job is failing. https://source.denx.de/u-boot/custodians/u-boot-microblaze/-/pipelines
Thanks, Michal
collected 1152 items / 1151 deselected / 1 selected test/py/tests/test_trace.py F [100%] =================================== FAILURES =================================== __________________________________ test_trace __________________________________ test/py/tests/test_trace.py:292: in test_trace check_function(cons, fname, proftool, map_fname, trace_dat) test/py/tests/test_trace.py:117: in check_function out = util.run_and_log(cons, ['sh', '-c', cmd]) test/py/u_boot_utils.py:181: in run_and_log output = runner.run(cmd, ignore_errors=ignore_errors, stdin=stdin, env=env) test/py/multiplexed_log.py:183: in run raise exception E ValueError: Exit code: 1 ---------------------------- Captured stdout setup ----------------------------- /u-boot trace: early enable at 00100000 sandbox_serial serial: pinctrl_select_state_full: uclass_get_device_by_phandle_id: err=-19 =>
I don't get it either since I see this on master with trace options enabled per CI: $ ./u-boot -T arch/sandbox/dts/test.dtb trace: early enable at 00100000 sandbox_serial serial: pinctrl_select_state_full: uclass_get_device_by_phandle_id: err=-19 ... rest of boot proceeds ...
So the test should be failing already if that was the problem.
I don't thin it is the problematic part.
When I look at https://source.denx.de/u-boot/u-boot/-/blob/master/tools/docker/Dockerfile?r...
clone is done from git clone https://github.com/rostedt/trace-cmd.git /tmp/trace/trace-cmd && \
https://github.com/rostedt/trace-cmd/blob/master/README#L5
we should be using git://git.kernel.org/pub/scm/utils/trace-cmd/trace-cmd.git instead.
I installed the latest version on my PC and run it via pytest and this is what I am getting. You see that trace-cmd report is returning "failed to init data".
Are we wired to any specific trace-cmd version?
Wait, no, what's going on exactly? The github repository is a mirror of the kernel.org one. So yes, we should correct to point to the main one, but the top of tree in both cases is commit 354dccca52e805ce1b22e2b62cbae8c265886c27.