[PATCH 2/2] Makefile: Build final mpc85xx non-SPL image in standard file u-boot.bin

[Pali Rohár]

+ Peng Fan

On Sunday 21 August 2022 08:17:29 Tom Rini wrote:
> On Wed, Aug 17, 2022 at 04:44:39PM -0600, Simon Glass wrote:
> > Hi,
> > 
> > On Wed, 17 Aug 2022 at 15:07, Pali Rohár <pali at kernel.org> wrote:
> > >
> > > On Wednesday 03 August 2022 13:28:01 Pali Rohár wrote:
> > > > On Monday 01 August 2022 19:15:46 Tom Rini wrote:
> > > > > On Mon, Aug 01, 2022 at 09:39:00PM +0200, Pali Rohár wrote:
> > > > > > On Monday 01 August 2022 13:13:22 Simon Glass wrote:
> > > > > > > Hi Pali,
> > > > > > >
> > > > > > > On Mon, 1 Aug 2022 at 09:43, Pali Rohár <pali at kernel.org> wrote:
> > > > > > > >
> > > > > > > > Currently Makefile produces final mpc85xx image when SPL is not used in
> > > > > > > > custom file u-boot-with-dtb.bin. It is quite confusing name as build
> > > > > > > > process produce also intermediate file standard file u-boot-dtb.bin (which
> > > > > > > > is just intermediate and not bootable). Other platforms use u-boot.bin
> > > > > > > > (UBOOT_BIN) as standard name for final bootable raw image.
> > > > > > > >
> > > > > > > > So change Makefile rules and binman to produce final bootable file for
> > > > > > > > mpc85xx also into file u-boot.bin. There is just need for mpc85xx to not
> > > > > > > > define default rule for u-boot.bin then instruct binman (via DTS file) to
> > > > > > > > store final image into u-boot.bin (instead of u-boot-with-dtb.bin) and
> > > > > > > > finally rename target u-boot-with-dtb.bin to u-boot.bin.
> > > > > > > >
> > > > > > > > With this change are also removed custom Makefile hacks for mpc85xx that it
> > > > > > > > produced non-standard output file. And also updated documentation.
> > > > > > > >
> > > > > > > > Signed-off-by: Pali Rohár <pali at kernel.org>
> > > > > > > > ---
> > > > > > > >  Makefile                             | 19 +++++--------------
> > > > > > > >  arch/powerpc/dts/kmcent2-u-boot.dtsi |  2 +-
> > > > > > > >  arch/powerpc/dts/u-boot.dtsi         |  2 +-
> > > > > > > >  board/freescale/p1_p2_rdb_pc/README  |  2 +-
> > > > > > > >  board/freescale/p2041rdb/README      |  3 ---
> > > > > > > >  board/freescale/t102xrdb/README      |  2 +-
> > > > > > > >  board/freescale/t104xrdb/README      |  2 +-
> > > > > > > >  board/freescale/t208xqds/README      |  2 +-
> > > > > > > >  board/freescale/t208xrdb/README      |  2 +-
> > > > > > > >  9 files changed, 12 insertions(+), 24 deletions(-)
> > > > > > >
> > > > > > > At present u-boot.bin has a very standard meaning - it is U-Boot with the DT.
> > > > > > >
> > > > > > > Boards which need something more than that can/should use binman to
> > > > > > > create a separate file.
> > > > > > >
> > > > > > > I certainly agree that u-boot-with-dtb.bin is a terrible name, though.
> > > > > > > Something more descriptive would be better.
> > > > > > >
> > > > > > > But is it possible to drop these SoC-specific rules in the Makefile
> > > > > > > and just build everything needed in the standard binman rule in the
> > > > > > > Makefile?
> > > > > > >
> > > > > > > Regards,
> > > > > > > Simon
> > > > > >
> > > > > > I do not know what is binman doing and how to use it. I just do not see
> > > > > > reason why it is needed to use such additional tool for building final
> > > > > > binary for powerpc/mpc85xx as other arm boards do not use it at all.
> > > > > >
> > > > > > Ad your comment "At present u-boot.bin has a very standard meaning - it
> > > > > > is U-Boot with the DT." - This is exactly what binman for mpc85xx
> > > > > > produces.
> > > > > >
> > > > > > So I see there could be improvements, but as a first step this my patch
> > > > > > should be enough?
> > > > >
> > > > > So, one of the issues with PowerPC stuff is that much of it is so far
> > > > > behind the rest of U-Boot in terms of frameworks.  So yes, let us start
> > > > > by fixing the functional problem you're describing here and then see
> > > > > what appetite exists for further work here.
> > > > >
> > > > > --
> > > > > Tom
> > > >
> > > > Ok, so these two patches in this patch series is a starting point.
> > > >
> > > > Now I send another patch which does another cleanup in this area:
> > > > https://patchwork.ozlabs.org/project/uboot/patch/20220803112442.4735-1-pali@kernel.org/
> > >
> > > PING?
> > 
> > I'm not sure what to say here and will leave it to Tom.
> 
> And I'm assuming that Marek will pick this up, along with all of your
> other outstanding PowerPC patches that are essentially related to making
> the original turris platform work.
> 
> -- 
> Tom