Any objections / concerns for the following change? An example. This rule in devfs.rules: add path da* mode 660 group operator and this directory: /data result in the following rule being actually installed: 100 path data group operator mode 660 Of course, I could refine the pattern in the rule, but I shouldn't have to do it, because the pattern is for /dev/ entries, not arbitrary files in the filesystem namespace. commit 7ce5e9ca5c107e2669f18efa472c1ab14999247c Author: Andriy Gapon <avg_at_icyb.net.ua> Date: Sat Mar 23 10:29:39 2013 +0200 rc.subr: disabling globbing while processing devfs rules in devfs_rulesets_from_file() The rules themselves typically have shell-like patterns and it is incorrect when they get replaced with matching filesystem entries. Shell magic by: jilles diff --git a/etc/rc.subr b/etc/rc.subr index f37ede7..9952c82 100644 --- a/etc/rc.subr +++ b/etc/rc.subr _at__at_ -1301,7 +1301,7 _at__at_ make_symlink() # devfs_rulesets_from_file() { - local file _err _me + local file _err _me _opts file="$1" _me="devfs_rulesets_from_file" _err=0 _at__at_ -1314,6 +1314,11 _at__at_ devfs_rulesets_from_file() debug "$_me: no such file ($file)" return 0 fi + + # Disable globbing so that the rule patterns are not expanded + # by accident with matching filesystem entries. + _opts=$-; set -f + debug "reading rulesets from file ($file)" { while read line do _at__at_ -1360,6 +1365,7 _at__at_ devfs_rulesets_from_file() break fi done } < $file + case $_opts in *f*) ;; *) set +f ;; esac return $_err } -- Andriy GaponReceived on Sat Mar 23 2013 - 07:41:56 UTC
This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:40:36 UTC