touch-packages team mailing list archive

[FEI <fei@xxxxxxxxxxxx>]

Public bug reported:

when using "normal exit 0 SIGTERM" in a job, and if the job is killed with SIGTERM, a warning message will appear in dmesg, like
[330159.038611] init: xxxyyyzzz main process (570) killed by TERM signal

If using "normal exit 0 SIGTERM 15" instead, there won't be such warning
message when the job is killed with SIGTERM, 15 is the numerical value
for SIGTERM, per "man 7 signal"

After reading the code, it looks to me that the issue is:
1. when parsing "normal exit 0 SIGTERM", the numerical value of SIGTERM is left shifted by 8 and then stored in job->class->normalexit
http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/parse_job.c#L2433
2. but when determining the appropriate log level, the value in job->class->normalexit is compared directly to numerical value of the signal, (without left shift the signal value)
http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/job_process.c#L1495

When using "normal exit 0 SIGTERM 15", both "15 << 8" and 15 will be stored in job->class->normalexit, the code to store 15 is at
http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/parse_job.c#L2428
and thus later on the code to determine log level works.

Note that "normal exit 0 SIGTERM" works well with "respawn", it is just
the logging code has the minor bug said above.

** Affects: upstart (Ubuntu)
     Importance: Undecided
         Status: New

-- 
You received this bug notification because you are a member of Ubuntu
Touch seeded packages, which is subscribed to upstart in Ubuntu.
https://bugs.launchpad.net/bugs/1515060

Title:
  Wrong logging level when signal name (e.g., SIGTERM) is used with
  normal exit stanza

Status in upstart package in Ubuntu:
  New

Bug description:
  when using "normal exit 0 SIGTERM" in a job, and if the job is killed with SIGTERM, a warning message will appear in dmesg, like
  [330159.038611] init: xxxyyyzzz main process (570) killed by TERM signal

  If using "normal exit 0 SIGTERM 15" instead, there won't be such
  warning message when the job is killed with SIGTERM, 15 is the
  numerical value for SIGTERM, per "man 7 signal"

  After reading the code, it looks to me that the issue is:
  1. when parsing "normal exit 0 SIGTERM", the numerical value of SIGTERM is left shifted by 8 and then stored in job->class->normalexit
  http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/parse_job.c#L2433
  2. but when determining the appropriate log level, the value in job->class->normalexit is compared directly to numerical value of the signal, (without left shift the signal value)
  http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/job_process.c#L1495

  When using "normal exit 0 SIGTERM 15", both "15 << 8" and 15 will be stored in job->class->normalexit, the code to store 15 is at
  http://bazaar.launchpad.net/~upstart-devel/upstart/trunk/view/head:/init/parse_job.c#L2428
  and thus later on the code to determine log level works.

  Note that "normal exit 0 SIGTERM" works well with "respawn", it is
  just the logging code has the minor bug said above.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/upstart/+bug/1515060/+subscriptions