[arch-projects] [initscripts] [PATCH 05/12] fsck_all, kill_everything: Clean up run_hook positions

Tom Gundersen teg at jklm.no
Sat Jun 25 15:11:19 EDT 2011



Sent from my iPhone

On Jun 25, 2011, at 20:15, Dave Reisner <d at falconindy.com> wrote:

> On Sat, Jun 25, 2011 at 12:15:57PM +0200, Kurt J. Bosch wrote:
>> * run postfsck hook within stat block as for prefsck
>> * run prekillall hook out of stat blocks as for postkillall
>>  (revert obsolete e39ec61b7d642b36368d84f240b96eeda3c43b2f)
>> ---
>> functions |    9 +++++----
>> 1 files changed, 5 insertions(+), 4 deletions(-)
>> 
>> diff --git a/functions b/functions
>> index 9c881c1..eab2074 100644
>> --- a/functions
>> +++ b/functions
>> @@ -283,8 +283,9 @@ kill_everything() {
>>    done
>> 
>>    # Terminate all processes
>> +    run_hook "${1}_prekillall"
>> +    
>>    stat_busy "Sending SIGTERM To Processes"
>> -    run_hook "$1_prekillall"
>>        local pid k5args=""
>>        for pid in ${omit_pids[@]}; do
>>            k5args+=" -o $pid"
>> @@ -298,7 +299,7 @@ kill_everything() {
>>        /bin/sleep 1
>>    stat_done
>> 
>> -    run_hook "$1_postkillall"
>> +    run_hook "${1}_postkillall"
>> }
> 
> variables can't start with a number, so the extra bracing isn't
> necessary.
> 
>> 
>> # Start/trigger UDev, load MODULES and settle UDev
>> @@ -378,13 +379,13 @@ fsck_all() {
>>        [[ -f /forcefsck ]] || in_array forcefsck $(< /proc/cmdline) && FORCEFSCK="-- -f"
>>        run_hook sysinit_prefsck
>>        fsck -A -T -C$FSCK_FD -a -t "$NETFS,noopts=_netdev" $FORCEFSCK >$FSCK_OUT 2>$FSCK_ERR
>> -    local fsckret=$?
>> +        local fsckret=$?
>> +        run_hook sysinit_postfsck
>>    if (( fsckret > 1 )); then
>>        stat_fail
>>    else
>>        stat_done
>>    fi
>> -    run_hook sysinit_postfsck
>>    return $fsckret
>> }
>> 
>> -- 
>> 1.7.1
>> 
> 
> Not a fan of this. I'd rather let the pass/fail come through before user
> hooks are run. There's potential for insanity, should a user decide to
> muck with the $fsckret value.
> 
> Please try not to mix whitespace with code changes, as well.
> 
> d

+1


More information about the arch-projects mailing list