phc2sys process exits with status 255

[Douglas Henrique Koerich]

Brief Description
-----------------
phc2sys service terminates with exit code 255 (-1) without any indication of root cause; other PTP services (ts2phc and ptp4l) continue to work without issues

Severity
--------
Major: feature is usable but degraded

Steps to Reproduce
------------------
- Stop the phc2sys service that is using interface "ifaceX" as master clock (-s ifaceX)
- Launch phc2sys program with same configuration (-f, -s, etc.) with debugging messages (-l 7)
- Provoke an instability in the interface "ifaceX" (e.g. iplink set dev ifaceX down/up)
- Wait for program to exit with code=255 and check user.log contains the message "ioctl PTP_SYS_OFFSET_EXTENDED: Device or resource busy" (it does not happen all the times, may need to repeat the instability step)

Expected Behavior
------------------
phc2sys service/program should continue to work

Actual Behavior
----------------
phc2sys service/program terminates with exit code=255

Reproducibility
---------------
Reproducible, but not 100% of times

System Configuration
--------------------
AIO-DX + worker(s)

Branch/Pull Time/Commit
-----------------------
stx6.0

Last Pass
---------
This scenario has not been tested before.

Timestamp/Logs
--------------
Original logs that started analysis:

2022-07-05T02:18:28.803 Notice: 2022-07-05 02:18:28 +0000 /Stage[main]/Platform::Ptpinstance/Ptp_config_files[phc-inst1]/Exec[enable-phc-inst1]/returns: executed successfully
2022-07-05T02:33:10.718 worker-0 systemd[1]: notice <email address hidden>: main process exited, code=exited, status=255/n/a
2022-07-05T02:33:10.745 worker-0 kernel: err [ 311.937905] ice 0000:51:00.0: PTP failed to get time
2022-07-05T02:33:13.000 worker-0 ts2phc: err [319.844] nmea source timed out

Logs from investigation (with steps to reproduce):

worker-0 phc2sys: debug [31190.453] phc-test ioctl PTP_SYS_OFFSET_EXTENDED: Device or resource busy

Test Activity
-------------
Evaluation

Workaround
----------
By removing "255" from phcsys's systemd configuration the service can at least restart

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.opendev.org/c/starlingx/integ/+/851437

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.opendev.org/c/starlingx/integ/+/851437
Committed: https://opendev.org/starlingx/integ/commit/0a9b0570e9266d7d03e17de85f67fb64041dee5d
Submitter: "Zuul (22348)"
Branch: master

commit 0a9b0570e9266d7d03e17de85f67fb64041dee5d
Author: Douglas Henrique Koerich <email address hidden>
Date: Thu Jul 28 13:42:53 2022 -0300

    Don't exit when reading of PHC fails with EBUSY

    Pull in the following upstream commits for both CentOS and Debian:
    https://sourceforge.net/p/linuxptp/code/ci/7824b13db9533ddebe37cf444d7aaa5d235575d3/
    https://sourceforge.net/p/linuxptp/code/ci/270709323a161ff1cb83af511ce50691152c75cf/
    https://sourceforge.net/p/linuxptp/code/ci/dadd2593c7beaee9eba5828a7bd8a0b5849dd8bb/
    https://sourceforge.net/p/linuxptp/code/ci/e8dc364f9fd5fbdac5d2c5e433f28e9da0028d49/
    See the patches themselves for more details.

    Closes-Bug: 1983022
    Signed-off-by: Douglas Henrique Koerich <email address hidden>
    Change-Id: I20f18b17752588b5b031eed88b5e52fb15c830fa