Patchwork [RFC,2/7] image.bbclass: run postinst scripts once, at rootfs time

This patch will allow the repeating postinst scriptlets to be run
only once, on host, at do_rootfs time. This will lower the time for
rootfs generation and, also, instead of running some time consuming
scriptlets at target's first boot, we will do on the host.

Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
---
 meta/classes/image.bbclass |   24 +++++++++++++++++++++++-
 1 file changed, 23 insertions(+), 1 deletion(-)

On 19 September 2012 12:49, Laurentiu Palcu <laurentiu.palcu@intel.com> wrote:
> +ROOTFS_POSTPROCESS_COMMAND ?= "run_intercept_scriptlets"

From other examples, it looks like this should be "run_intercept_scriptlets;"

Also whilst this is a great proof of concept, maybe it should be
integrated into image.bblcass directly instead of being a postprocess
hook?  A postprocess hook might expect the postinstall scripts to have
been ran already, which they would have been if they were not
intercepted.

Ross

On 09/19/2012 06:14 PM, Burton, Ross wrote:
> On 19 September 2012 12:49, Laurentiu Palcu <laurentiu.palcu@intel.com> wrote:
>> +ROOTFS_POSTPROCESS_COMMAND ?= "run_intercept_scriptlets"
> 
> From other examples, it looks like this should be "run_intercept_scriptlets;"
Will do. My bad.

> 
> Also whilst this is a great proof of concept, maybe it should be
> integrated into image.bblcass directly instead of being a postprocess
> hook?  A postprocess hook might expect the postinstall scripts to have
> been ran already, which they would have been if they were not
> intercepted.
I'm a little confused here... Technically, when the postprocess hooks
are executed all the postinstall scripts were already executed. Ideally,
the run-once scripts are created from inside the postinst scriptlets
themselves. So, I don't really understand why using the postprocess hook
isn't safe. Can you please elaborate?

Thanks,
Laurentiu
> 
> Ross
>

On 19 September 2012 16:45, Laurentiu Palcu <laurentiu.palcu@intel.com> wrote:
> I'm a little confused here... Technically, when the postprocess hooks
> are executed all the postinstall scripts were already executed. Ideally,
> the run-once scripts are created from inside the postinst scriptlets
> themselves. So, I don't really understand why using the postprocess hook
> isn't safe. Can you please elaborate?

My concern (and it's just a hunch) is that there could be a
post-process hook running before the intercepts are run which could be
expecting the post install phase to have been completed.  With
intercepts a portion of the post-install phase is happening later than
they previously did, so they haven't completely been ran yet.

Ross

On 09/19/2012 06:50 PM, Burton, Ross wrote:
> On 19 September 2012 16:45, Laurentiu Palcu <laurentiu.palcu@intel.com> wrote:
>> I'm a little confused here... Technically, when the postprocess hooks
>> are executed all the postinstall scripts were already executed. Ideally,
>> the run-once scripts are created from inside the postinst scriptlets
>> themselves. So, I don't really understand why using the postprocess hook
>> isn't safe. Can you please elaborate?
> 
> My concern (and it's just a hunch) is that there could be a
> post-process hook running before the intercepts are run which could be
> expecting the post install phase to have been completed.  With
> intercepts a portion of the post-install phase is happening later than
> they previously did, so they haven't completely been ran yet.
I see your point now... The only place we could do it in image.bbclass
is immediately after rootfs_${IMAGE_PKGTYPE}_do_rootfs in do_rootfs()
function. But then we fall into the same issue you mentioned. The
post-process hooks would be executed before the intercept scripts...

The alternative would be to use the post-install hooks
(ROOTFS_POSTINSTALL_COMMAND) instead of post-process. These are executed
before the post-process hooks.

Thanks,
Laurentiu

> 
> Ross
>

On Wed, 2012-09-19 at 14:49 +0300, Laurentiu Palcu wrote:
> This patch will allow the repeating postinst scriptlets to be run
> only once, on host, at do_rootfs time. This will lower the time for
> rootfs generation and, also, instead of running some time consuming
> scriptlets at target's first boot, we will do on the host.
> 
> Signed-off-by: Laurentiu Palcu <laurentiu.palcu@intel.com>
> ---
>  meta/classes/image.bbclass |   24 +++++++++++++++++++++++-
>  1 file changed, 23 insertions(+), 1 deletion(-)
> 
> diff --git a/meta/classes/image.bbclass b/meta/classes/image.bbclass
> index ab212b3..4889260 100644
> --- a/meta/classes/image.bbclass
> +++ b/meta/classes/image.bbclass
> @@ -155,7 +155,7 @@ inherit ${IMAGE_CLASSES}
>  
>  IMAGE_POSTPROCESS_COMMAND ?= ""
>  MACHINE_POSTPROCESS_COMMAND ?= ""
> -ROOTFS_POSTPROCESS_COMMAND ?= ""
> +ROOTFS_POSTPROCESS_COMMAND ?= "run_intercept_scriptlets"
>  
>  # some default locales
>  IMAGE_LINGUAS ?= "de-de fr-fr en-gb"
> @@ -174,6 +174,21 @@ do_build[nostamp] = "1"
>  # task, so that we have a single fakeroot context for the whole process.
>  do_rootfs[umask] = "022"
>  
> +
> +run_intercept_scriptlets () {
> +	echo $D >> test.txt
> +	if [ -d ${WORKDIR}/intercept_scripts ]; then
> +		cd ${WORKDIR}/intercept_scripts
> +		echo "Running intercept scripts:"
> +		for script in *; do
> +			if [ "$script" = "*" ]; then break; fi
> +			echo "> Executing $script"
> +			chmod +x $script
> +			./$script
> +		done
> +	fi
> +}
> +
>  fakeroot do_rootfs () {
>  	#set -x
>  	# When use the rpm incremental image generation, don't remove the rootfs
> @@ -202,6 +217,13 @@ fakeroot do_rootfs () {
>  		done
>  	fi
>  
> +	if [ -d ${WORKDIR}/intercept_scripts ]; then
> +		# clean-up old intercept scripts
> +		rm -f ${WORKDIR}/intercept_scripts/*
> +	else
> +		mkdir ${WORKDIR}/intercept_scripts
> +	fi

You can achieve this piece with:

do_rootfs[cleandirs] += "${WORKDIR}/intercept_scripts"

Cheers,

Richard