[oe] packaged staging issue; additional packages

[Richard Purdie]

On Sat, 2010-01-30 at 13:18 +0100, Frans Meulenbroeks wrote:
> I seem to have found an issue with packaged staging.
> I'll try to describe the problem, but maybe my understanding is not
> 100% so bear with me in that case
[...]
> I've focused myself on this one:
> 
>  * opkg_install_cmd: Cannot install package udev.
>  * satisfy_dependencies_for: Cannot satisfy the following dependencies
> for sysvinit:
>  * satisfy_dependencies_for:    sysvinit-inittab *
>  * opkg_install_cmd: Cannot install package sysvinit.
> 
> What seems to be the case here is the following:
> sysvinit bb file contains the following:
> 
> RDEPENDS_${PN} = "${PN}-inittab"
> 
> PACKAGES =+ "bootlogd ${PN}-inittab"
> FILES_bootlogd = "/etc/init.d/bootlogd /etc/init.d/stop-bootlogd
> /etc/rc?.d/S*bootlogd /sbin/bootlogd"
> FILES_${PN}-inittab = "${sysconfdir}/inittab"
> 
> So it generates the sysvinit-inittab package and RDEPENDS on it.
> 
> If I search for inittab in my work/sysvinit dir I get:
> ./inittab
> ./sysvinit-2.86/man/inittab.5
> ./staging-pkg/staging/armv7a-angstrom-linux-gnueabi/usr/share/man/man5/inittab.5
> ./staging-pkg/pkgdata/beagleboard-angstrom-linux-gnueabi/runtime/sysvinit-inittab
> ./staging-pkg/pkgdata/beagleboard-angstrom-linux-gnueabi/runtime/sysvinit-inittab.packaged
> ./sysroot-destdir/home/frans/oe/tmp_angstrom/staging/armv7a-angstrom-linux-gnueabi/usr/share/man/man5/inittab.5
> ./packages-split/sysvinit-inittab
> ./packages-split/sysvinit-inittab/etc
> ./packages-split/sysvinit-inittab/etc/inittab
> ./packages-split/sysvinit-doc/usr/share/man/man5/inittab.5
> ./package/etc/inittab
> ./package/usr/share/man/man5/inittab.5
> ./image/etc/inittab
> ./image/usr/share/man/man5/inittab.5
> 
> So the file inittab does not end up in packaged staging.
> 
> It seems to me that packaged-staging.bbclass needs some work and that
> something like staging_packager needs to be involved for all packages
> in a recipe.
> Howeever, implementing this is beyond my python capabilities and also
> beyond my understanding of the build system.
> 
> Is my analysis correct?
> Can someone implement something or provide some guidance?

inittab is a false lead but you're heading in the right direction. The
staging package should contain a sysvinit-inittab*.ipk. Whilst I only
have a Poky build handy, I see:

find -name sysvinit-inittab*
./packages-split/sysvinit-inittab
./staging-pkg/deploy/ipk/netbook/sysvinit-inittab_2.86-r32_netbook.ipk
./staging-pkg/pkgdata/netbook-poky-linux/runtime/sysvinit-inittab
./staging-pkg/pkgdata/netbook-poky-linux/runtime/sysvinit-inittab.packaged

The key line being:

./staging-pkg/deploy/ipk/netbook/sysvinit-inittab_2.86-r32_netbook.ipk

If something like that file isn't there, something in OE's package
staging is broken.

Cheers,

Richard