On Thu, Oct 30, 2008, Derrick Karpo wrote:
> Just to recount and answer some of Michael's questions, none of the
> acpid modules exist in this test as they are all compiled in. If all
> acpi modules are compiled into the kernel then the acpid fails to
> start at boot. If they are left as modules then acpid successfully
> starts at boot. With "2>/dev/null" removed there is no more useful
> information on boot but manually running the script post-boot will
> show the modprobe warnings that the modules don't exist.
Aha, thanks, I think I got it:
Around module-init-tools-3.3-pre11/modprobe.c:1373, modprobe will
either error("") and return or if unknown_silent is set will exit(1).
unknown_silent is set with modprobe -q/-Q/--quiet/--silent which is set
via MODPROBE_OPTIONS in the acpid init script. This is only set if
VERBOSE is false.
So there's a different behavior in terms of exit status between:
modprobe --all foobar (exit 0)
modprobe --all --quiet foobar (exit 1)
This is quite problematic as --all is supposed to proceed and attempt
to load all modules, but in --quiet it stops as soon as a module can
not be loaded...
So I think I'll drop --quiet for now and report the bug against
modprobe.
--
Loïc Minier
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
