[OE-core] [PATCH 2/2] nativesdk-*-provides-dummy: Fixes to allow correct operation with opkg

Richard Purdie richard.purdie at linuxfoundation.org
Mon Feb 11 14:45:06 UTC 2019


On Mon, 2019-02-11 at 06:35 -0800, Khem Raj wrote:
> On Wed, Feb 6, 2019 at 6:39 PM Phil Reid <preid at electromag.com.au>
> wrote:
> > On 3/01/2019 9:59 pm, Richard Purdie wrote:
> > > On Mon, 2018-12-31 at 20:56 +0100, Andrej Valek wrote:
> > > > Your previous solution is working for me, so I was not working
> > > > on
> > > > fixing
> > > > this. I thought, that You will just add missing packages, and
> > > > that
> > > > is.
> > > > I don't have setup for failing machine.
> > > > 
> > > > Is there any simple way to get all missing perl packages?
> > > 
> > > I reworked the patch a bit, tested and just merged a working
> > > version of
> > > it:
> > > 
> > > http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=d4e27863f3874fb7dc543eed8dcf6a8c4d146fb5
> > > 
> > > Cheers,
> > > 
> > > Richard
> > > 
> > 
> > This patch seems to trigger the following new warning for me during
> > populate_sdk
> > 
> > Collected errors:
> >   * Solver encountered 1 problem(s):
> >   * Problem 1/1:
> >   *   - package busybox-dev-1.29.2-r0.4.cortexa9t2hf-neon requires
> > busybox = 1.29.2-r0.4, but none of the providers can be installed
> >   *
> >   * Solution 1:
> >   *   - allow deinstallation of target-sdk-provides-dummy-1.0-
> > r0.9.sdk-provides-dummy-target
> > 
> >   *   - do not ask to install a package providing target-sdk-
> > provides-dummy-dev
> > 
> >   * Solution 2:
> >   *   - do not ask to install a package providing busybox-dev
> > 
> > 
> > Not sure what is request busybox-dev to be istalled in the sdk.
> > Any thoughts?
> > 
> 
> Richard
> 
> I am seeing same failure as well. I solved it by adding
> 
> RDEPENDS_${PN}-dev = "" in busybox recipe.
> 
> I am not sure if thats right solution, busybox-dev seems to be empty
> package.

Its not. The real question is why busybox is being installed into the
SDK (which in turn pulls in busybox-dev). I don't think it is being
pulled in on our test images so its probably both your local
configurations that trigger something we don't see on the autobuilder.
Perhaps something is missing from target-sdk-provides-dummy.bb?

I'm also guessing this is thud so its possible we're missing a backport
of a related patch...

Cheers,

Richard



More information about the Openembedded-core mailing list