[Pkg-acpi-devel] Bug#502704: Bug#502704: Bug#502704: Bug#502704: acpid is

[Derrick Karpo]

On Fri, Oct 24, 2008 at 2:57 AM, Michael Meskes <meskes at debian.org> wrote:
> On Fri, Oct 24, 2008 at 10:19:55AM +0200, Loïc Minier wrote:
>> On Fri, Oct 24, 2008, Michael Meskes wrote:
>> > if [ $MODULES ]; then
>> >     modprobe --all --use-blacklist $MODULES 2>/dev/null
>> > fi
>>
>>  Err you probably lack a -n here, but FYI there's already:
>
> No. For some strange reasons the "if [ $MODULES ]" works for me as does
> "if [ ! -z $MODULES ]" but "if [ -n $MODULES ]" does not.
>
>>     if [ -z "$MODULES" ]; then
>>         return
>>     fi
>
> Ah sorry, missed this.
>
>>  I think modprobe --all always returns 0 (which is probably a bug; I've
>>  reported when I rewrote the init script), so it might be a bogus return
>>  again which might want to be return 0, but I don't think so.
>
> Probably not. From what I understood the modprobe is executed on his system, so
> the return is not.
>
> Derrick, could you please add an echo and output $MODULES on your system with
> some boundaries, so we see whether it's empty or maybe containing some empty
> string or whatever? Also could you send us the modules.dep file?
>
> Michael
>

My previous tests confirmed that $MODULES is not empty and contains
well defined space-delimited modules as defined in /etc/default/acpid.
 I've also confirmed that modules.dep checks out so that shouldn't be
the issue but it was worth checking out.

In looking at the /etc/init.d/acpid script from Ubuntu it appears to
solve many of the modprobe issues that I am seeing.  It checks to
ensure the modules exist before loading them and won't attempt to load
modules twice.  Perhaps it's worth merging chunks of their
load_modules() into the Debian script.

Derrick