[OE-core] [RFC] kernel: Enable externalsrc on kernels which instantiate kernel.bbclass

[Bruce Ashfield]

On Wed, Dec 3, 2014 at 9:36 AM, Paul Eggleton
<paul.eggleton at linux.intel.com> wrote:
> On Wednesday 03 December 2014 08:14:44 Bruce Ashfield wrote:
>> On Wed, Dec 3, 2014 at 7:23 AM, Laurentiu Palcu
>>
>> <laurentiu.palcu at intel.com> wrote:
>> > Hi Paul,
>> >
>> > On Wed, Dec 03, 2014 at 12:00:31PM +0000, Paul Eggleton wrote:
>> >> On Monday 22 September 2014 13:04:47 Bruce Ashfield wrote:
>> >> > On 14-09-22 01:03 PM, Khem Raj wrote:
>> >> > > On Mon, Sep 22, 2014 at 8:27 AM, Bruce Ashfield
>> >> > >
>> >> > > <bruce.ashfield at windriver.com> wrote:
>> >> > >> But the reports we've been getting have been that externalsrc
>> >> > >> builds are working for kernels, and linux-yocto without the change
>> >> > >> in place, so I'm looking to reduce the patch footprint and
>> >> > >> re-submit.
>> >> > >
>> >> > > no, it cant work if folks were trying the usecase I have mentioned.
>> >> > > The fix is infact for
>> >> > > non linux-yocto kernels.
>> >> >
>> >> > I'll send out my updated patch shortly, it should fix both cases.
>> >>
>> >> Bruce, did you get around to doing this?
>> >>
>> >> At the moment I am trying to make my workflow tool ("devtool modify")
>> >> work with linux-yocto. I've at least got to the point where I can
>> >> extract the appropriate source from the recipe, but when I use
>> >> externalsrc to point to it and then try to build it, it doesn't even get
>> >> past do_configure, so something must be missing (make complains about
>> >> missing targets, presumably because it's running in ${B} and there's
>> >> nothing in that directory).
>> >>
>> >> Do you expect linux-yocto + externalsrc to work at the moment in master?
>> >> If
>> >> so, are there any special steps I need to do with respect to preparing
>> >> the
>> >> external source tree that my tool might not be performing?
>> >
>> > FWIW, I'm using externalsrc to build non linux-yocto kernels but, to make
>> > it work, I had to overwrite the KERNEL_CONFIG_COMMAND:
>> >
>> > KERNEL_CONFIG_COMMAND = "oe_runmake_call -C ${S} O=${B} oldnoconfig || yes
>> > '' | oe_runmake -C ${S} O=${B} oldconfig"
>>
>> For an externalsrc kernel, we've always skipped configure* completely.
>> If the user is providing the kernel, it should also be configured. We don't
>> want the build system changing anything at all,  including the config that is
>> in place.
>
> OK, that makes sense - but all I'm doing is enabling externalsrc for the
> kernel, and do_configure is still being executed. Perhaps that's something that
> shouldn't be happening (needs to be added to SRCTREECOVEREDTASKS? We don't
> normally cover do_configure for other recipes.)

IMHO it should be, I have it inhibited here.

Still trying to get my 1.8 work cleaned up for a RFC, and I'll send
you what I have for
this shortly as well.

Bruce

>
> Cheers,
> Paul
>
> --
>
> Paul Eggleton
> Intel Open Source Technology Centre



-- 
"Thou shalt not follow the NULL pointer, for chaos and madness await
thee at its end"