On Tuesday 10 February 2009 12:54:31 Ben Warren wrote: > Mike Frysinger wrote: > > We use things from autoconf.mk in the top level Makefile, so it needs to > > exist before doing a top level make. Otherwise, the top level make env > > does not get the values in the autoconf.mk. > > > > Signed-off-by: Mike Frysinger <vap...@gentoo.org> > > --- > > Makefile | 2 ++ > > 1 files changed, 2 insertions(+), 0 deletions(-) > > > > diff --git a/Makefile b/Makefile > > index 787c5f2..db77d3b 100644 > > --- a/Makefile > > +++ b/Makefile > > @@ -3293,6 +3293,8 @@ BFIN_BOARDS = bf533-ezkit bf533-stamp bf537-stamp > > bf561-ezkit > > > > $(BFIN_BOARDS:%=%_config) : unconfig > > @$(MKCONFIG) $(@:_config=) blackfin blackfin $(@:_config=) > > + @$(MAKE) -s -B $(obj)include/autoconf.mk > > + @$(MAKE) -s -B $(obj)include/autoconf.mk > > Do you really mean to do this twice?
unfortunately, yes. since some settings in the board config are turned into compiler flags and those compiler flags can in turn affect the board config, we need to do it twice. first is to make sure the proper cpu flags are propagated into the toplevel build env while the second is to make sure the autoconf.mk fully reflects the board config. i guess i could add a one line comment above each one giving hints about why each is needed ... -mike
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
