[OE-core] [PATCH 5/6] kernel.bbclass: remove references to update-modules

Darren Hart dvhart at linux.intel.com
Wed Jan 23 16:24:27 UTC 2013



On 01/23/2013 12:22 AM, Laurentiu Palcu wrote:
> 
> 
> On 01/22/2013 10:48 PM, Darren Hart wrote:
>>> -    use_update_modules = oe.utils.contains('DISTRO_FEATURES', 'update-modules', True, False, d)
>>>> -    if use_update_modules:
>>>> -        postinst = d.getVar('pkg_postinst_modules', True)
>>>> -        postrm = d.getVar('pkg_postrm_modules', True)
>>>> -    else:
>>>> -        postinst = None
>>>> -        postrm = None
>>>> +    postinst = d.getVar('pkg_postinst_modules', True)
>>>> +    postrm = d.getVar('pkg_postrm_modules', True)
>> This seems to be inverted logic from the original. If update-modules is
>> removed, then use_update_modules should be false right? Which would have
>> previously set postinst and postrm to None.
> I believe both the previous and current logic are right:
>   * Before, the update-modules script was called in the postinst/postrm.
> So, when the update-modules package was not part of the build, the
> postinst/postrm could not call update-modules anymore. As a consequence,
> they were set to none.
>   * After update-modules has been removed, we are not constrained by
> calling update-modules anymore in postinst/postrm but we can safely call
> depmod to have the dependencies computed.
> 
> So, is there anything wrong with the current logic?

Ah, I see. That makes sense to me.

Thanks,

-- 
Darren Hart
Intel Open Source Technology Center
Yocto Project - Technical Lead - Linux Kernel




More information about the Openembedded-core mailing list