[Buildroot] [PATCH 1/1] package/dhcp: use internal bind

[Fabrice Fontaine]

Le mer. 6 avr. 2022 à 15:30, Fabrice Fontaine
<fontaine.fabrice at gmail.com> a écrit :
>
> Le mer. 6 avr. 2022 à 14:58, Tim Hammer <Tim.Hammer at orolia.com> a écrit :
> >
> > On Wed, Apr 06, 2022 at 09:08:58AM +0200, Fabrice Fontaine wrote:
> > > Le mer. 6 avr. 2022 à 03:58, Tim Hammer <Tim.Hammer at orolia.com> a écrit :
> > > >
> > > > On Tue, Apr 05, 2022 at 06:18:15PM +0200, Peter Seiderer wrote:
> > > > > Hello Tim,
> > > > >
> > > > > On Tue, 5 Apr 2022 12:15:14 +0000, Tim Hammer <Tim.Hammer at orolia.com> wrote:
> > > > >
> > > > > > On Tue, Apr 05, 2022 at 08:19:08AM +0200, Fabrice Fontaine wrote:
> > > > > > > Dear Tim,
> > > > > > >
> > > > > > > Le mar. 5 avr. 2022 à 05:54, Tim Hammer <Tim.Hammer at orolia.com> a écrit :
> > > > > > > >
> > > > > > > > On Sun, Apr 03, 2022 at 12:03:18PM +0200, Fabrice Fontaine wrote:
> > > > > > > > > Use internal bind as dhcp doesn't build since bump of bind to version
> > > > > > > > > 9.16.26 in commit 8adeaec8afacbc680edebae91524f6144926ac92 and upstream
> > > > > > > > > doesn't plan to fix it any time soon:
> > > > > > > > > https://gitlab.isc.org/isc-projects/dhcp/-/issues/233#note_276883
> > > > > > > > >
> > ...
> > > > > >
> > > > > >
> > > > > > > However, I assume that BUILD_CC and BUILD_CFLAGS in bind.mk should
> > > > > > > also be updated?
> > > > > >
> > > > > > I do not have the issue building bind (9.16.26), there lib/dns/Makefile.in seems to be better written:
> > > > > >
> > > > > >      gen: gen.c
> > > > > >     ${HOSTCC} ${HOST_CFLAGS} -I${top_srcdir}/lib/isc/include \
> > > > > >     ${HOST_LDFLAGS} -o $@ ${srcdir}/gen.c
> > > > > >
> > > > > > Perhaps we should be patching that file instead of the config?
> > > > >
> > > > > See https://git.buildroot.net/buildroot/tree/package/bind/0001-cross.patch
> > > > >
> > > > > Regards,
> > > > > Peter
> > > > >
> > > >
> > > > Yes, that reinforces my thinking. Unfortunately, the bundled bind is not unzipped until the build step
> > > > and I have not figured out a good way to get this diff in place.
> > > > Perhaps someone with more experience can see the solution that is eluding me?
> > >
> > > My bad, I made a typo in my first mail, you should use $(HOSTCC) and
> > > not $(HOST_CC).
> >
> > Wow, how many times did I look at those lines and not see the difference! ;-(
> > Thanks for catching that!
> >
> > > Patching bind is not needed.
> > > I can send the patch or let you handle it.
> >
> > I am still concerned about my ability to verify-
> > I added "corrections" for BUILD_CFLAGS & BUILD_LDFLAGS so those match the changes in bind.mk,
> >         -DHCP_BIND_EXTRA_CONFIG = BUILD_CC='$(TARGET_CC)'
> >         +
> >         +DHCP_BIND_EXTRA_CONFIG = BUILD_CC='$(HOSTCC)'
> >         +DHCP_BIND_EXTRA_CONFIG += BUILD_CFLAGS='$(HOST_CFLAGS)'
> >         +DHCP_BIND_EXTRA_CONFIG += BUILD_LDFLAGS='$(HOST_LDFLAGS)'
> > but I am unsure about the others used (BUILD_CPPFLAGS, BUILD_LIBS, LFS_*)...
> >         ${BUILD_CC} ${BUILD_CFLAGS} -I${top_srcdir}/lib/isc/include \
> >         ${LFS_CFLAGS} ${LFS_LDFLAGS} \
> >         ${BUILD_CPPFLAGS} ${BUILD_LDFLAGS} -o $@ ${srcdir}/gen.c \
> >         ${BUILD_LIBS} ${LFS_LIBS}
> > Should those be set to empty strings? Or is it safe to assume they are undefined by default?
>
> BUILD_CPPFLAGS should be set to HOST_CPPFLAGS.
> IMHO it is safe to assume that BUILD_LIBS and LFS_* are empty.
>
> Feel free to send your patch with a link to the autobuilder failure such as:
>
> Fixes:
>  - http://autobuild.buildroot.org/results/7a5cdf30881d208807976cf98960c5fe2abfed50

The correct link is
http://autobuild.buildroot.org/results/da6fd904d1a6bae73b6ff89dd008de1f459bb7d7

The previous one is about zlib, I'll investigate it.


>
> Arnout or the others will make a second review and spot any remaining issues.
>
> >
> > >
> > ...
> > >
> > > Best Regards,
> > >
> > > Fabrice
> >
> > .Tim
> > Tim D. Hammer
> > tim.hammer at orolia.com
> >
> >
>
> Best Regards,
>
> Fabrice

Best Regards,

Fabrice