[Bug 575154] Re: dashism in checkroot/checkfs/usplash functions

Steve Langasek steve.langasek at canonical.com
Fri May 17 21:57:16 UTC 2013


as noted, this bug no longer exists in Ubuntu.

** Changed in: sysvinit (Ubuntu)
       Status: New => Invalid

-- 
You received this bug notification because you are a member of Ubuntu
Foundations Bugs, which is subscribed to sysvinit in Ubuntu.
https://bugs.launchpad.net/bugs/575154

Title:
  dashism in checkroot/checkfs/usplash functions

Status in “sysvinit” package in Ubuntu:
  Invalid

Bug description:
  Binary package hint: sysvinit

  When trying to run Hardy with mksh as /bin/sh booting fails. I now managed to track it down
  to a bug in initscripts still present in maverick:
  https://launchpad.net/ubuntu/+archive/primary/+files/sysvinit_2.87dsf-4ubuntu17.dsc

  Let me illustrate the actual problem:

  tglase at tglase:~ $ mksh -c 'sleep 2 & sleep 1; wait %1; echo $?'
  0
  tglase at tglase:~ $ mksh -c 'sleep 1 & sleep 2; wait %1; echo $?'
  127
  tglase at tglase:~ $ dash -c 'sleep 1 & sleep 2; wait %1; echo $?'
  0
  tglase at tglase:~ $ bash -c 'sleep 1 & sleep 2; wait %1; echo $?'
  bash: line 0: wait: %1: no such job
  127

  Basically, the “wait” builtin returns 127 if the job has already finished by the time
  wait is called, except in dash, where it returns 0. This makes *buntu unusable
  with a shell other than dash as /bin/sh (among other things). I also wonder
  whether this is the cause (with not using mksh as /bin/sh though) for our
  workstations (at work) to boot to multiuser with / in read-only mode even
  when it wants the 30-mount check (we need to manually fsck and reboot
  then, which is idiotic in a production / company environment).

  I’m currently working on a fix, will attach that soonish.

To manage notifications about this bug go to:
https://bugs.launchpad.net/ubuntu/+source/sysvinit/+bug/575154/+subscriptions




More information about the foundations-bugs mailing list