[Bug 960079] Re: race condition in USR1 signal handling in mountall

[Eelco Dolstra]

** Patch added: "install the USR1 signal handler before daemonising"
   https://bugs.launchpad.net/bugs/960079/+attachment/2902857/+files/fix-usr1-race.patch

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to mountall in Ubuntu.
https://bugs.launchpad.net/bugs/960079

Title:
  race condition in USR1 signal handling in mountall

Status in “mountall” package in Ubuntu:
  New

Bug description:
  There is a slight race condition in the way mountall installs its USR1
  signal handler. It does that after daemonising, so as a result the
  mountall-net job might send it the USR1 signal after the mountall job
  has reached the start/running state but before it has installed the
  signal handler.  This will result in mountall getting killed.

  Example of an automated test run showing this problem:
  http://hydra.nixos.org/build/2309161/log/raw.  The relevant line is:

  client2# [   10.963784] init: mountall main process (1971) killed by
  USR1 signal

  The attached patch (to mountall 2.35) fixes this by moving the
  installation of the USR1 handler to before the fork.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/mountall/+bug/960079/+subscriptions