[PATCH]unify mainboard makefiles

[Patrick Georgi]

Hi,

after moving crt0s and lscripts away from the mainboards' Makefiles,
those Makefiles are rather simple now, with more opportunities to move
out things.

Attached patch does:
- make HAVE_HARD_RESET match what newconfig did
- introduce BOARD_HAS_HARD_RESET and use it if a board provides
hard_reset in $(MAINBOARDDIR)/reset.c, instead of some chipset component
- move a couple of rules out of the mainboards' Makefiles into
src/arch/i386/Makefile.inc:
    initobj-y += crt0.o
    obj-y += mainboard.o
    obj-$(CONFIG_GENERATE_MP_TABLE) += mptable.o
    obj-$(CONFIG_GENERATE_PIRQ_TABLE) += irq_tables.o
    obj-$(CONFIG_BOARD_HAS_HARD_RESET) += reset.o
- remove Makefile.incs that are empty (or comment-only) after these
changes, incl. Makefile.romccboard.inc (and references to it)
- Make include not fail if file doesn't exist.

reset.o used to be HAVE_HARD_RESET, but to make this rule more
universal, that new variable had to be introduced.

The rationale for this work is that it unifies the mainboard layout,
which leads to less cargo cult when creating a new board, which we
usually start by copying from a similar one.
It also makes it easier to do global changes, as more common
functionality is in common files.

It's build tested and has no regression vs. current HEAD (ie. the same
two boards fail for the same reason) and is boot tested on kontron/986lcd-m.

Signed-off-by: Patrick Georgi

Future work in this area: Move ROMCCFLAGS into Kconfig or global
Makefiles, same for the ACPI related rules. Many more
mainboard-Makefiles can be removed then.

Patrick

[Myles Watson]

On Fri, Feb 26, 2010 at 3:38 PM, Patrick Georgi wrote:
> Hi,
>
> after moving crt0s and lscripts away from the mainboards' Makefiles,
> those Makefiles are rather simple now, with more opportunities to move
> out things.
>
> Attached patch does:
> make HAVE_HARD_RESET match what newconfig did
> introduce BOARD_HAS_HARD_RESET and use it if a board provides
>
> hard_reset in $(MAINBOARDDIR)/reset.c, instead of some chipset component
> move a couple of rules out of the mainboards' Makefiles into
>
> src/arch/i386/Makefile.inc:
>        initobj-y += crt0.o
>        obj-y += mainboard.o
>        obj-$(CONFIG_GENERATE_MP_TABLE) += mptable.o
>        obj-$(CONFIG_GENERATE_PIRQ_TABLE) += irq_tables.o
>        obj-$(CONFIG_BOARD_HAS_HARD_RESET) += reset.o
> remove Makefile.incs that are empty (or comment-only) after these
>
> changes, incl. Makefile.romccboard.inc (and references to it)
> Make include not fail if file doesn't exist.
>
> reset.o used to be HAVE_HARD_RESET, but to make this rule more
> universal, that new variable had to be introduced.

Can we put HAVE_HARD_RESET into the ck804 Kconfig instead of into all the
ck804 mainboards?
> It's build tested and has no regression vs. current HEAD (ie. the same
> two boards fail for the same reason) and is boot tested on
> kontron/986lcd-m.
>
> Signed-off-by: Patrick Georgi

Acked-by: Myles Watson

Thanks,
Myles

[Peter Stuge]

Patrick Georgi wrote:
> Signed-off-by: Patrick Georgi

Acked-by: Peter Stuge
> Future work in this area: Move ROMCCFLAGS into Kconfig or global
> Makefiles, same for the ACPI related rules.

I'm tempted to say Kconfig, though I'm not sure I will like the
outcome.
> Many more mainboard-Makefiles can be removed then.

Awesome progress, this!

//Peter

[Stefan Reinauer]

On 2/26/10 11:38 PM, Patrick Georgi wrote:
> config HAVE_HARD_RESET
>    bool
> default y if BOARD_HAS_HARD_RESET
> default n
> help
>  This variable specifies whether a given board has a hard_reset
>  function, no matter if it's provided by board code or chipset code.
>
> +config BOARD_HAS_HARD_RESET
> bool
> default n
> help
>  This variable specifies whether a given board has a reset.c
>  file containing a hard_reset() function.

Hm this is kind of semi-beautiful, but I think it's good for getting the
job done... Fight them mainboard makefiles!

Acked-by: Stefan Reinauer

Happy weekend everyone!

Stefan

[Peter Stuge]

Stefan Reinauer wrote:
> > +config BOARD_HAS_HARD_RESET
>
> Hm this is kind of semi-beautiful, but I think it's good for
> getting the job done... Fight them mainboard makefiles!

The alternative would be to move all reset functions out of board
code into chipset code. Maybe that's actually feasible?

//Peter

[Stefan Reinauer]

On 2/27/10 1:19 AM, Peter Stuge wrote:
> Stefan Reinauer wrote:
> > > +config BOARD_HAS_HARD_RESET
> >
> > Hm this is kind of semi-beautiful, but I think it's good for
> > getting the job done... Fight them mainboard makefiles!
>
> The alternative would be to move all reset functions out of board
> code into chipset code. Maybe that's actually feasible?

Yes, I start believing reset code belongs into the southbridge, because
the southbridge decodes cf9.
But that's good for another day. Patrick has done incredible work on
simplifying coreboot, let's go one step at once.

Stefan

[Peter Stuge]

Stefan Reinauer wrote:
> > The alternative would be to move all reset functions out of board
> > code into chipset code. Maybe that's actually feasible?
>
> Yes, I start believing reset code belongs into the southbridge,
> because the southbridge decodes cf9.

Makes sense.
> But that's good for another day. Patrick has done incredible work
> on simplifying coreboot,

Yes indeed.
> let's go one step at once.

Fair. :)

//Peter