On Wed, Jun 12, 2013 at 02:39:13PM +0200, Heiko Schocher wrote:
Hello Tom,
Am 12.06.2013 14:28, schrieb Tom Rini:
On Wed, Jun 12, 2013 at 10:56:01AM +0200, Heiko Schocher wrote:
Hello tom,
your
commit 4596dcc1d4ea5763e0f92cf5becd9fc7d4c6e674 Author: Tom Rini trini@ti.com Date: Fri May 31 12:31:59 2013 -0400
am33xx/omap: Move save_omap_boot_params to omap-common/boot-common.cintroduced, that all am335x based boards must call save_omap_boot_params() from the board specific s_init function. I just stumbled over it, as I updated to current head and my upcoming am335x based siemens boards didn't boot anymore ... should we think about to move this s_init() to a common place, and extract board specific things? Maybe arch/arm/cpu/armv7/omap-common/boot-common.c is a place for it?
I think the first non-am335x_evm board pulled the code we need in board/ a bit too far in the board-specific direction.
Whacking the WDT (which could be done differently I imagine based on your other patches), calling save_omap_boot_params and UART stuff is common. Figuring out which DDR and how is not. I think we can learn from omap-common/hwinit-common.c::s_init but need to have our own in arch/arm/cpu/armv7/am33xx/board.c and declared __weak or wrapped with CONFIG_AM33XX since TI814x (and TI816x) are different. Or maybe some more thinking share still.
Do you have time for this? Thanks!
I try to find some ;-)
Maybe something like this:
add in arch/arm/cpu/armv7/am33xx/board.c a weak s_init(), which all am35xx boards use, and call in this s_init() a s_init_board() which all am335x boards must have defined?
s/s_init_board/sdram_init/ and yes. If we whack the UART stuff out into uart_enable (ala ti814x) and add a board_enable_early_pinmux (for the muxes needed, and put this right after save_omap_...) we might not need to make s_init __weak.