Re: [PATCHv2] {kernel, module}.bbclass: don't run depmod for module packages during do_rootfs

[Andreas Oberritter <obi@opendreambox.org>]

On 25.04.2012 03:14, Darren Hart wrote:
> 
> 
> On 04/24/2012 05:42 PM, Andreas Oberritter wrote:
>> On 25.04.2012 02:29, Darren Hart wrote:
>>>
>>>
>>> On 04/23/2012 02:48 PM, Andreas Oberritter wrote:
>>>> * depmod already gets executed by pkg_postinst_kernel-image.
>>>>
>>>> * If you build a module using module.bbclass,
>>>>   pkg_postinst returns 1 in do_rootfs, causing
>>>>   pkg_postinst to run again on first boot. To
>>>>   improve this situation, I copied pkg_postinst
>>>>   from kernel.bbclass to module.bbclass. This was
>>>>   rejected by Koen, because he doesn't like the
>>>>   code from kernel.bblcass, which uses
>>>>   ${STAGING_DIR_KERNEL}. Richard then suggested
>>>>   that calling depmod during do_rootfs wasn't
>>>>   necessary at all, because it already gets done by
>>>>   kernel-image.
>>>>
>>>
>>> Thanks for adding that in. I'm fine not addressing the reliance on the
>>> existence of $D for now (no worse than it was).
>>
>> Can you explain what could be improved?'
> 
> I did in the previous thread:
> 
> http://lists.linuxtogo.org/pipermail/openembedded-core/2012-April/021419.html

Sorry, I must have missed the inline comments when I replied to that mail.

From the manpage of test(1):

       -d FILE
              FILE exists and is a directory

       -z STRING
              the length of STRING is zero

I'm not sure what using -d should accomplish. (Virtually?) all postinst
scripts in OE test for $D's emptiness to decide whether they are called
offline (during do_rootfs) or online (on the target), so if a package
manager sets $D, it should rather stop doing that, instead of relying on
$D not being a directory by chance.

Regards,
Andreas