Re: [PATCH 8/8] linux-yocto-dev: update to v4.15+

["Wold, Saul" <saul.wold@intel.com>]

On Thu, 2017-12-07 at 05:30 -0500, Bruce Ashfield wrote:
> 
> 
> On Wed, Dec 6, 2017 at 11:28 AM, Burton, Ross <ross.burton@intel.com>
> wrote:
> > The amended commit has mismatching commit log and patch (target
> > elfutils is in the commit log but not the patch), and doesn't
> > mention installing target elfutils if they want it.
> > 
> 
> 
> I fixed up the commit log on that patch to match the actual code.
> 
> As it turns out, I don't see the need to install target elfutils
> anymore, so I didn't
> include anything about that in the log.
> 
> We still have some lttng build issues with the -dev kernel (due to
> objtool being
> required and the openssl headers required for making the scripts),
> and those
> are being worked on separately. Having this updated linux-yocto-dev
> recipe
> in the tree gets the kernel building and booting and enables us to
> work through
> the other issues more quickly (not to meention, I've already pushed
> 4.15-rc2 to
> the actual kernel tree, so you get that from a build regardless).
> 
With Regards to the objtool, I was also seeing and issue with kernel-
devsrc, and objtool/.debug showing up in the devsrc packages-split.  I
am trying to get a solid set of steps that reproduces this.  The
strange thing was I did not see the .debug files in the kernel build
area or even in the kernel-devsrc/image directory but it was strangely
showed up in packages-split and failed a QA check.

Sau!

> Cheers,
> 
> Bruc
> 
> 
>  
> > Ross
> > 
> > On 4 December 2017 at 17:43, Bruce Ashfield <bruce.ashfield@windriv
> > er.com> wrote:
> > > On 2017-12-04 12:39 PM, Richard Purdie wrote:
> > > > On Mon, 2017-12-04 at 12:24 -0500, Bruce Ashfield wrote:
> > > > > On 2017-12-04 11:38 AM, Khem Raj wrote:
> > > > > > On Mon, Dec 4, 2017 at 7:39 AM, Bruce Ashfield
> > > > > > <bruce.ashfield@windriver.com> wrote:
> > > > > > > Outside of the normal patch refreshes and boot issues,
> > > > > > > there are
> > > > > > > new
> > > > > > > build time tools within the kernel that required the
> > > > > > > following
> > > > > > > dependencies:
> > > > > > > 
> > > > > > > For ORC_UNWINDER support in x86-64:
> > > > > > > 
> > > > > > >    DEPENDS += "${@bb.utils.contains('ARCH', 'x86',
> > > > > > > 'elfutils-
> > > > > > > native elfutils', '', d)}"
> > > > > > > 
> > > > > >  do we need both target and host elfutils
> > > > > > 
> > > > >  Yup. There were references to both. Some had to run for
> > > > > hostcc
> > > > > and others in the target arch.
> > > > > 
> > > >  
> > > > Just for reference this is pretty bad for performance as it
> > > > delays the
> > > > kernel compile until some substantial parts of userspace build.
> > > > 
> > > > 
> > >  
> > > On a second look, I can likely turn off the target part, if
> > > someone
> > > wants it, they can always install the package (or it could be a
> > > rdepends).
> > > 
> > > I'll amend the commit and leave it on that branch with just the
> > > DEPENDS on the -native version.
> > > 
> > > > Is ORC_UNWINDER useful and commonly used?
> > > > 
> > >  
> > > The upstream kernel commit turned it on by default, I turned it
> > > off in
> > > the kernel-cache, but I wanted to make sure the dependency was
> > > in place.
> > > 
> > > The commit series from Josh Poimboeuf all lead to it being on as
> > > the default choice (even with a slight overhead).
> > > 
> > > Bruce
> > > 
> > > 
> > > > Cheers,
> > > > 
> > > > Richard
> > > > 
> > > > 
> > >  
> > > -- 
> > > _______________________________________________
> > > Openembedded-core mailing list
> > > Openembedded-core@lists.openembedded.org
> > > http://lists.openembedded.org/mailman/listinfo/openembedded-core
> > 
> > 
> > --
> > _______________________________________________
> > Openembedded-core mailing list
> > Openembedded-core@lists.openembedded.org
> > http://lists.openembedded.org/mailman/listinfo/openembedded-core
> > 
> 
> 
>