[OE-core] slightly confused about .bbappend file and FILESPATH

Robert P. J. Day rpjday at crashcourse.ca
Tue Dec 11 12:54:52 UTC 2012


On Tue, 11 Dec 2012, Martin Jansa wrote:

> On Thu, Dec 06, 2012 at 01:19:57PM -0500, Robert P. J. Day wrote:
> > On Thu, 6 Dec 2012, Mark Hatle wrote:
> >
> > > On 12/6/12 12:03 PM, Robert P. J. Day wrote:
> > > >
> > > >    currently writing a tutorial page on how .bbappend files and
> > > > FILESPATH work, and i wanted to use a live example, but now i'm
> > > > confused.
> > >
> > > The way it works is the same way as PATH in the shell.  It's just a set of
> > > search paths.
> > >
> > > >    using meta-ti layer, i chose to build core-image-minimal for a
> > > > beagleboard, for which the meta-ti layer has the following under
> > > > recipes-core/netbase/ that i want to use as an example:
> > > >
> > > > $ find
> > > > .
> > > > ./netbase_5.0.bbappend
> > > > ./netbase-5.0
> > > > ./netbase-5.0/beagleboard
> > > > ./netbase-5.0/beagleboard/interfaces
> > > > $
> > > >
> > > > where the .bbappend file contains only:
> > > >
> > > > THISDIR := "${@os.path.dirname(bb.data.getVar('FILE', d, True))}"
> > >
> > > I believe this is the default value of 'THISDIR' already, so there is no
> > > reason to specify it in the .bbappend itself.
> > >
> > > > FILESPATH =. "${@base_set_filespath(["${THISDIR}/${PN}"], d)}:"
> > >
> > > In this example, it's incorrect.  As you have a directory named
> > > "netbase-5.0", so your FILESPATH should be: ${THISDIR}/${PN}-${PV}
> >
> >   ok, so that's just a bug in the setting of FILESPATH in that one
> > example?  whew.  i was seriously wondering how badly i misunderstood
> > how this feature worked.
> >
> >   as soon as my beagle core-image-minimal finishes building, i should
> > be able to trivially check that the interfaces file used to populate
> > the rootfs image is the one from oe-core and not the one that should
> > have come from meta-ti.
>
> See FILESEXTRAPATHS in other bbappends and
>
> meta/classes/base.bbclass:FILESPATH = "${@base_set_filespath(["${FILE_DIRNAME}/${BP}","${FILE_DIRNAME}/${BPN}", "${FILE_DIRNAME}/files"], d)}"
>
> + definition of fce base_set_filespath:
> meta/classes/utils.bbclass:def base_set_filespath(path, d):

  i'm not sure what i'm supposed to be seeing in the above.  are you
saying that the above will automatically handle the example i posted
above?

  i configured the meta-ti layer and used chris larson's bitbake-env
utility (which should really be added as an official tool to bitbake
given its awesome awesomeness) and examined FILESPATH for the netbase
recipe:

$ bitbake-env -r netbase FILESPATH | tr : '\n'
Parsing recipes..done.
# FILESPATH="${@base_set_filespath(["${THISDIR}/${PN}"], d)}
${@base_set_filespath(["${FILE_DIRNAME}/${BP}",
"${FILE_DIRNAME}/${BPN}", "${FILE_DIRNAME}/files"], d)}"
FILESPATH="/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/linux-gnueabi
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/arm
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/build-linux
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/pn-netbase
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/dm814x-evm
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/ti814x
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/armv7a
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/class-target
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/forcevariable
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/libc-glibc
/home/rpjday/OE/dist/layers/meta-ti/recipes-core/netbase/netbase/
/home/rpjday/OE/dist/layers/oe-core/meta/recipes-core/netbase/netbase-5.0/linux-gnueabi
... snip ...

  as you can see, all of the checking done in the meta-ti layer
represents the netbase/netbase directory, which won't help since, in
meta-ti, it's actually netbase/netbase-5.0/.  i built a dm8148 image
using meta-ti and, sure enough, the network interfaces file that
should have come from meta-ti came instead from oe-core for what i
concluded was a fairly obvious reason.

  or am i misundertanding something?  wouldn't be the first time...

rday

-- 

========================================================================
Robert P. J. Day                                 Ottawa, Ontario, CANADA
                        http://crashcourse.ca

Twitter:                                       http://twitter.com/rpjday
LinkedIn:                               http://ca.linkedin.com/in/rpjday
========================================================================




More information about the Openembedded-core mailing list