[OE-core] Error while building qemu without X11 support.

Robert Yang liezhi.yang at windriver.com
Tue Sep 2 06:28:52 UTC 2014



On 09/02/2014 02:25 PM, Richard Purdie wrote:
> On Tue, 2014-09-02 at 14:21 +0800, Robert Yang wrote:
>>
>> On 09/02/2014 02:17 PM, Richard Purdie wrote:
>>> On Tue, 2014-09-02 at 00:33 -0300, Otavio Salvador wrote:
>>>> On Tue, Sep 2, 2014 at 12:30 AM, Robert Yang <liezhi.yang at windriver.com> wrote:
>>>>>
>>>>>
>>>>> On 09/02/2014 11:17 AM, Otavio Salvador wrote:
>>>>>>
>>>>>> On Mon, Sep 1, 2014 at 11:51 PM, Robert Yang <liezhi.yang at windriver.com>
>>>>>> wrote:
>>>>>>>
>>>>>>>
>>>>>>> On 09/02/2014 10:30 AM, Otavio Salvador wrote:
>>>>>>>>
>>>>>>>>
>>>>>>>> Hello,
>>>>>>>>
>>>>>>>> During our regression tests we found the following error:
>>>>>>>>
>>>>>>>> directfb fsl-image-machine-test at imx6qsabresd (1/4) consider defining a
>>>>>>>> PREFERRED_PROVIDER entry to match jpeg
>>>>>>>> directfb fsl-image-machine-test at imx6qsabresd (1/4) ERROR: Nothing
>>>>>>>> PROVIDES 'libxext-native' (but
>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> Did you use meta/recipes-graphics/xorg-lib/libxext_git.bb ? This one
>>>>>>> doesn't provide native, but libxext_1.3.2.bb does.
>>>>>>>
>>>>>>> Seems that we need remove the git version since it is older:
>>>>>>>
>>>>>>> PV = "1.0.99.1+gitr${SRCPV}"
>>>>>>>
>>>>>>> And doesn't see a license issue.
>>>>>>
>>>>>>
>>>>>> This has been reproduced in our autobuilder.
>>>>>>
>>>>>> It removes:
>>>>>>
>>>>>> x11 and wayland
>>>>>>
>>>>>> and adds:
>>>>>>
>>>>>> directfb
>>>>>>
>>>>>> Can you try to reproduce it?
>>>>>>
>>>>>
>>>>> Yes, I can reproduce the error, sorry, please revert this as a quick
>>>>> workaround:
>>>>>
>>>>> Author: Robert Yang <liezhi.yang at windriver.com>
>>>>> Date:   Tue Jul 29 01:02:44 2014 -0700
>>>>>
>>>>>       qemu-native: depends on libxext-native when enable sdl
>>>>>
>>>>>
>>>>> The DISTRO_FEATURES affects the native recipe, I'm not sure whether this is
>>>>> intended or not.
>>>>
>>>> This is an error in our autobuilder for regression test; Richard,
>>>> could you revert this in OE-Core?
>>>
>>> I think that illustrates the problem but is perhaps not the correct fix.
>>>
>>> We did hit some issues like this with poky-tiny and I ended up fixing
>>> this like:
>>>
>>> http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=8b70d4b2863f48e843fb6ac9e4a939ae63338093
>>>
>>> and also
>>>
>>> http://git.yoctoproject.org/cgit.cgi/poky/commit/?id=fd0398f2c1355597a95242e6c8400eae6ad60fa4
>>>
>>
>> Hello, how about this RFC patch:
>>
>> Subject: [RFC PATCH] distro_features_check.bbclass: skip for native and nativesdk
>>
>> The distro features should have no effect to the native or nativesdk
>> recipe, for example, if we need build a tool for target based on
>> directfb, the native recipe should still be built based on x11 which is
>> more common on the host.
>>
>> Signed-off-by: Robert Yang <liezhi.yang at windriver.com>
>> ---
>>    meta/classes/distro_features_check.bbclass |    4 ++++
>>    1 file changed, 4 insertions(+)
>>
>> diff --git a/meta/classes/distro_features_check.bbclass
>> b/meta/classes/distro_features_check.bbclass
>> index 1f1d6fb..16a194e 100644
>> --- a/meta/classes/distro_features_check.bbclass
>> +++ b/meta/classes/distro_features_check.bbclass
>> @@ -8,6 +8,10 @@
>>    # Copyright 2013 (C) O.S. Systems Software LTDA.
>>
>>    python () {
>> +    # DISTRO_FEATURES should not affect the native or nativesdk
>> +    if bb.data.inherits_class('native', d) or
>> bb.data.inherits_class('nativesdk', d):
>> +        return
>> +
>>        required_distro_features = d.getVar('REQUIRED_DISTRO_FEATURES', True)
>>        if required_distro_features:
>>            required_distro_features = required_distro_features.split()
>
> No, this will hack around half the problem and cause issues since
> DISTRO_FEATURES still does not contain x11 in the native case. The
> better solution is as I mention, correctly allow x11 in DISTRO_FEATURES
> for native. For nativesdk, you'd probably make a choice if your SDK was
> to include x11.

Thanks, got it.

// Robert

>
> Cheers,
>
> Richard
>
>
>
>



More information about the Openembedded-core mailing list