Re: [PATCH] kbuild: let object files depend on recordmcount instead of its sources

["Uwe Kleine-König" <u.kleine-koenig@pengutronix.de>]

Hello,

On Fri, Oct 07, 2011 at 03:19:21PM +0200, Uwe Kleine-König wrote:
> This has the upside that distributions only have to ship the recordmcount
> binary to allow building modules instead of recordmcount and it's sources.
> 
> Make still notices that the object files need to be rebuilt when
> recordmcount's sources change.
> 
> Signed-off-by: Uwe Kleine-König <u.kleine-koenig@pengutronix.de>
> ---
>  scripts/Makefile.build |    9 ++++-----
>  1 files changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/scripts/Makefile.build b/scripts/Makefile.build
> index a0fd502..bc83e94 100644
> --- a/scripts/Makefile.build
> +++ b/scripts/Makefile.build
> @@ -271,8 +271,7 @@ sub_cmd_record_mcount =					\
>  	if [ $(@) != "scripts/mod/empty.o" ]; then	\
>  		$(objtree)/scripts/recordmcount $(RECORDMCOUNT_FLAGS) "$(@)";	\
>  	fi;
> -recordmcount_source := $(srctree)/scripts/recordmcount.c \
> -		    $(srctree)/scripts/recordmcount.h
> +recordmcount_exec := $(objtree)/scripts/recordmcount
>  else
>  sub_cmd_record_mcount = set -e ; perl $(srctree)/scripts/recordmcount.pl "$(ARCH)" \
>  	"$(if $(CONFIG_CPU_BIG_ENDIAN),big,little)" \
> @@ -280,7 +279,7 @@ sub_cmd_record_mcount = set -e ; perl $(srctree)/scripts/recordmcount.pl "$(ARCH
>  	"$(OBJDUMP)" "$(OBJCOPY)" "$(CC) $(KBUILD_CFLAGS)" \
>  	"$(LD)" "$(NM)" "$(RM)" "$(MV)" \
>  	"$(if $(part-of-module),1,0)" "$(@)";
> -recordmcount_source := $(srctree)/scripts/recordmcount.pl
> +recordmcount_exec := $(srctree)/scripts/recordmcount.pl
>  endif
>  cmd_record_mcount = 						\
>  	if [ "$(findstring -pg,$(_c_flags))" = "-pg" ]; then	\
> @@ -301,13 +300,13 @@ define rule_cc_o_c
>  endef
>  
>  # Built-in and composite module parts
> -$(obj)/%.o: $(src)/%.c $(recordmcount_source) FORCE
> +$(obj)/%.o: $(src)/%.c $(recordmcount_exec) FORCE
>  	$(call cmd,force_checksrc)
>  	$(call if_changed_rule,cc_o_c)
I could reproduce the build failure Steve is seeing now (enabling
CONFIG_FUNCTION_TRACER helps, stupid me). I think the problem is that
with my change empty.o depends on recordmcount and in a way that I don't
understand recordmcount depends on empty.o. I wonder if it's possible to
get rid of the latter dependency?!

Best regards
Uwe

-- 
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | http://www.pengutronix.de/  |