[OE-core] [PATCH 1/2] V2 Incremental image generation(rpm based rootfs)

Robert Yang liezhi.yang at windriver.com
Tue Jan 10 10:40:19 UTC 2012


Hi Richard,

Please see my comments inline ...

On 01/06/2012 11:22 PM, Richard Purdie wrote:
> On Sat, 2011-12-31 at 16:10 +0800, Robert Yang wrote:
>> Incremental image generation, the rootfs would be totally removed and
>> re-created in the second generation by default, but with INC_IMAGE_GEN =
>> "1", the rootfs would be kept, and will do update(remove/add some pkgs)
>> on it.
>> it.
>>
>> This only for the rpm based rootfs, I don't know whether we also need
>> this for ipk or deb  based rootfs.
>
> As Koen points out, this should be called INC_RPM_IMAGE_GEN since it
> only works for rpm at present. It would be good to make this work on the
> other package backends. If we keep the generic variable name, the other
> package backends could error out saying they don't support it.
>

I did a rough investigation on deb these days, it seems not easy to do in
M2, so I'd like to make it work for both deb and ipk in M3. I will change the
name to INC_RPM_IMAGE_GEN at present, and send a V3 sooner.

>> [YOCTO #1651]
>>
>> Signed-off-by: Robert Yang<liezhi.yang at windriver.com>
>> ---

>> +
>> +        # Attempt to remove unwanted pkgs, the scripts(pre, post, etc.) has not
>> +        # been run by now, so don't have to run them(preun, postun, etc.) when
>> +        # erase the pkg
>
> Really? I'd have thought it possible the scripts have run during the
> previous do_rootfs?
>

Yes, I think so, I think these post scripts are put in /etc/rpm-postinsts/,
and will be run when the system starts at the first time. The best solution
is that remove the script when "rpm -e", but it is hard to know which pkg
owns which script, so I just prevent running the preun and postun when
"rpm -e", I think this is acceptable since the INC_RPM_IMAGE_GEN is not
suggested in the productive rootfs.

// Robert

> Otherwise the patch looks reasonable to me.
>
> Cheers,
>
> Richard
>
>
>




More information about the Openembedded-core mailing list