Re: [PATCH 1/5] {kernel, module}.bbclass: don't run depmod for module packages during do_rootfs

[Darren Hart <dvhart@linux.intel.com>]

On 04/23/2012 02:21 PM, Andreas Oberritter wrote:
> On 23.04.2012 19:35, Darren Hart wrote:
>>
>>
>> On 04/17/2012 06:05 AM, Andreas Oberritter wrote:
>>> * depmod already gets executed by pkg_postinst_kernel-image.
>>>
>>
>> Please include a description of the errors seen that this fixes. This
>> makes it easier for people experiencing this to find the fix.
> 
> Well, there is no error. 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. I already explained this in the cover letter and in the
> previous email thread.

Apologies if I missed some context. However, the detail needs to be in
the commit log, not the cover letter, as the commit log is where
developers will look for the rationale and justification of a change.

--
Darren

> 
>>
>>> Signed-off-by: Andreas Oberritter <obi@opendreambox.org>
>>> ---
>>>  meta/classes/kernel.bbclass |    4 +---
>>>  meta/classes/module.bbclass |    7 +++----
>>>  2 files changed, 4 insertions(+), 7 deletions(-)
>>>
>>> diff --git a/meta/classes/kernel.bbclass b/meta/classes/kernel.bbclass
>>> index 3519e7c..c21ab96 100644
>>> --- a/meta/classes/kernel.bbclass
>>> +++ b/meta/classes/kernel.bbclass
>>> @@ -276,9 +276,7 @@ fi
>>>  }
>>>  
>>>  pkg_postinst_modules () {
>>> -if [ -n "$D" ]; then
>>> -	${HOST_PREFIX}depmod -A -b $D -F ${STAGING_KERNEL_DIR}/System.map-${KERNEL_VERSION} ${KERNEL_VERSION}
>>> -else
>>> +if [ -z "$D" ]; then
>>
>> I understand it parallels the previous test - but it seems fragile to
>> rely on the mere existence of $D. Consider the following?
>>
>> if [ ! -d "$D" ]; then
>>
>> This will do the right thing if D is empty as well as if $D is defined
>> by bitbake.
>>
>>>  	depmod -a
>>>  	update-modules || true
>>>  fi
>>> diff --git a/meta/classes/module.bbclass b/meta/classes/module.bbclass
>>> index 53c16b7..91628e4 100644
>>> --- a/meta/classes/module.bbclass
>>> +++ b/meta/classes/module.bbclass
>>> @@ -37,15 +37,14 @@ module_do_install() {
>>>  }
>>>  
>>>  pkg_postinst_append () {
>>> -	if [ -n "$D" ]; then
>>> -		exit 1
>>> -	fi
>>> +if [ -z "$D" ]; then
>>>  	depmod -a
>>>  	update-modules || true
>>> +fi
>>>  }
>>>  
>>>  pkg_postrm_append () {
>>> -	update-modules || true
>>> +update-modules || true
>>
>> Unintentional whitespace change?
>>
> 

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