[OE-core] [PATCH 2/2] quilt: fix test for target build

Khem Raj raj.khem at gmail.com
Fri Apr 15 14:22:38 UTC 2011


On Fri, Apr 15, 2011 at 2:19 AM, Qing He <qing.he at intel.com> wrote:
> On Thu, 2011-04-14 at 23:23 +0800, Khem Raj wrote:
>> > The reason to touch this part of code is that the test itself doesn't
>> > work for newer versions of GNU patch, see:
>> > ??http://lists.nongnu.org/archive/html/quilt-dev/2010-06/msg00002.html
>> > This surely fixes the configure error, but is logically not suitable for
>> > cross compiling.
>>
>> As I understand you are trying to fix quilt target recipe. Then may it
>> will be better to take the above
>> patch for both target and native but apply this
>> workaround conditionally only for target recipe
>
> This was once backported, but subsequently reverted, see:
>  http://git.pokylinux.org/cgit/cgit.cgi/poky/commit/?id=ec740ac770d2124cbae1ab936d85b8343076c53b
>
> autoconf-native is not available at the point of quilt-native, so
> autoreconf is not desirable. This is also the reason for separation
> of quilt and quilt-native from the very beginning.


So you regenerate configure and add that to patch too along with .ac
file and then it should not regenerate it.

>
> Apart from that, the workaround is solely for target recipe.
>
>>
>> there are options --with-patch --with-diff also
>> there to point these utilities
>
> The problem still lies in cross building:
>  --with-patch=${bindir}/patch          at build time, it's /usr/bin/patch
>                                        which is host utility, if target
>                                        ${bindir} /= /usr/bin, the file
>                                        may not exist.
>
>  --with-patch=${STAGING_BINDIR}/patch  causes runtime error as
>                                        ${STAGING_BINDIR} leaks into target scripts
>
>  --with-patch=patch                    this may work, but may fail (in case
>                                        the target scripts unset PATH)
>
> This is the problem of such tests without the concept of a target
> sysroot.
>


this is fine we can apply this conditionally for target recipe.

> Thanks,
> Qing
>




More information about the Openembedded-core mailing list