Bug#931719: 40-systemd: line 11: 1: unbound variable when sourcing /lib/lsb/init-functions with $1 unset

[Mert Dirik]

>> After upgrading to Debian 10 we noticed a slightly different behaviour
>> in  /lib/lsb/init-functions.d/40-systemd which caused one of our scripts
>> to break and which also causes some /etc/init.d files like
>> /etc/init.d/mysql
>> to no longer to display their help text when calling them without a
>> parameter.
>>
>> The root cause are accesses to the $1 and $2 variables without checking
>> if they are defined beforehand. When running code using "set -u", this
>> causes the above mentioned error.
>>

Which command did you use to make that happen? I'm asking because I
can't test it right now.

>> I've outlined this very simple fix which bypasses all further tests if
>> $1 and so on are undefined.
>>
>> 	$ diff -C 9 40-systemd.orig 40-systemd
>> 	*** 40-systemd.orig     2019-05-24 22:58:59.000000000 +0200
>> 	--- 40-systemd  2019-07-09 16:40:39.642678140 +0200
>> 	***************
>> 	*** 1,18 ****
>> 	--- 1,21 ----
>> 	  # -*-Shell-script-*-
>> 	  # /lib/lsb/init-functions
>>
>> 	  _use_systemctl=0
>> 	  if [ -d /run/systemd/system ]; then
>>
>> 	      if [ -n "${__init_d_script_name:-}" ]; then # scripts run
>> with new init-d-script
>> 		  executable="$__init_d_script_name"
>> 		  argument="$1"
>> 	+     elif [ $# -eq 0 ] ; then
>> 	+         executable="$0"
>> 	+         argument=""
>> 	      elif [ "${0##*/}" = "init-d-script" ] ||
>> 		   [ "${0##*/}" = "${1##*/}" ]; then # scripts run with
>> old  init-d-script
>> 		  executable="$1"
>> 		  argument="$2"
>> 	      else # plain old scripts
>> 		  executable="$0"
>> 		  argument="$1"
>> 	      fi
>
> Mert, since this basically your code, could you have a look at this.
>

It seems OK.

What I'd rather do is drop the "##*/" substitution, and set a default
value if $1 is unset, and let the statement continue:

         executable="$__init_d_script_name"
         argument="$1"
     elif [ "${0##*/}" = "init-d-script" ] ||
-         [ "${0##*/}" = "${1##*/}" ]; then # scripts run with old
init-d-script
+         [ "${0}" = "${1:-}" ]; then # scripts run with old  init-d-script
         executable="$1"
-        argument="$2"
+        argument="${2:-}"
     else # plain old scripts
         executable="$0"
         argument="$1"

Dropping the "##*/" substitution isn't actually related to this change
but dropping it allows the use of ":-" syntax in the same line. I was
in fact thinking about dropping it irrelevant to this issue.


Another important issue here is the use of "set -e" and "set -u" on
the /etc/init.d/mysql script. It is ineffective right now because
later in 40-systemd file they are set back to their defaults (set +e
and set +u). I don't know the policy or best practices about this
topic but either one of the scripts should be accommodated to be in
compliance with the other.

I'm sorry I can't test any of this right now as I don't have a
suitable and up-to-date test system ready and I'm not sure when I'll
be.

Regards