[oe] SRC_URI_arch override. error?

[Chris Larson]

On Tue, Sep 21, 2010 at 12:51 PM, Denys Dmytriyenko <denis at denix.org> wrote:

> On Wed, Sep 08, 2010 at 04:52:20PM +0200, Frans Meulenbroeks wrote:
> > Hi,
> >
> > I have an issue with the linux-libc-headers recipe for nios2.
> > Maybe I am trying to do something what is not possible.
> >
> > The recipe contains:
> > SRC_URI = "${KERNELORG_MIRROR}/pub/linux/kernel/v2.6/linux-${PV}.tar.bz2
> \
> >         "
> >
> > SRC_URI_nios2 =
> > "
> ftp://opensource.axon.nl/mirror/git_sopc.et.ntust.edu.tw.linux-2.6.git_a32ca88c4f3f3850c5c9789db2afab2530c6856d.tar.gz;name=nios2tarball
> > \
> >         "
> >
> > I was under the assumption that when building with TARGET_ARCH set to
> > "nios2" (e.g. for machine neek) I would get the 2nd uri.
> > However actually the system seems to concatenate these two URI's.
> >
> > I have locally modified classes/base.bbclass function base_do_unpack
> > do dump some info:
> > python base_do_unpack() {
> >     from glob import glob
> >
> >     srcurldata = bb.fetch.init(d.getVar("SRC_URI", True).split(), d,
> True)
> >     bb.note("unpacking SRC_URI %s" % d.getVar("SRC_URI", True))
> >     bb.note("unpacking SRC_URI split %s" % d.getVar("SRC_URI",
> True).split())
> >     bb.note("unpacking srcurldata %s" % srcurldata)
>
> Frans,
>
> Are you still facing this issue?
>
> Looks like I'm in the same boat with you on this one - I have an amended
> recipe in a local overlay, which slightly modifies the source tarball name
> in
> SRC_URI. It used to work fine before, but lately it tries to call
> do_unpack()
> on both - the source tarball with the name from amend.inc, as well as the
> source tarball with the name from the original SRC_URI. Since it doesn't do
> do_fetch() for both, only do_unpack(), it fails if the other tarball was
> not
> previously downloaded into DL_DIR...
>
> I was able to trace the problem down to the do_unpack() rewrite commit:
>
>
> http://cgit.openembedded.org/cgit.cgi/openembedded/commit/?id=900cc29b603691eb3a077cb660545ead3715ed54
>
> Before this commit, SRC_URI used to behave properly, even with
> modifications
> and amendments. I'm still trying to debug the new code to see where the
> problem happens.
>
> Maybe, Chris, as the author of that code, can spot the issue sooner.
> Thanks.


This thread has been sitting in my todo list for a while now, haven't had a
chance to dive in.  I'll see if I can take a look tonight.
-- 
Christopher Larson
clarson at kergoth dot com
Founder - BitBake, OpenEmbedded, OpenZaurus
Maintainer - Tslib
Senior Software Engineer, Mentor Graphics