Dear Reinhard Meyer,
Am 17.10.2010 um 20:06 schrieb Reinhard Meyer:
Dear Andreas Bießmann,
+Andreas Bie�mannandreas.devel@gmail.com
On my e-Mail client the sharp-s in the commit shows wrongly, however it does show ok in the e-mail header itself.
int board_init (void) {
- at91_pio_t *pio = (at91_pio_t *)AT91_PIO_BASE;
- /* Enable Ctrlc */
- console_init_f ();
- console_init_f();
I think that console_init_f is done in the arm/board.c already, please verify that.
Yes it is really. The call on this place is historic and caused no errors therefore I left it there. But it should be removed.
diff --git a/board/atmel/at91rm9200ek/config.mk b/board/atmel/at91rm9200ek/config.mk index 9ce161e..7b99e02 100644 --- a/board/atmel/at91rm9200ek/config.mk +++ b/board/atmel/at91rm9200ek/config.mk @@ -1 +1,2 @@ -TEXT_BASE = 0x21f00000 +# currently only NOR flash booting is supported +TEXT_BASE = 0x10000000
Currently TEXT_BASE is being changed to CONFIG_ and moved to the <board>.h file, thus removing the need for a <board>/config.mk.
Yes I know. But when submitting this patch the rework for boards.cfg was not upstream.
I'm not sure when to pick up this patch and how it will merge. Wolfgang?
I would prefer to get this patch in 2010.12, I could do the requested changes til monday evening. This board Is the basis for getting the messy at91rm9200 stuff cleaned up. I hope to get most drivers merged with current at91 code basis for 2011.03.
diff --git a/include/configs/at91rm9200ek.h b/include/configs/at91rm9200ek.h index 145c3c3..8e9db92 100644 --- a/include/configs/at91rm9200ek.h +++ b/include/configs/at91rm9200ek.h
...
/* ARM asynchronous clock */ /*
- from 18.432 MHz crystal
- (18432000 / 4 * 39)
*/ -#define AT91C_MAIN_CLOCK 179712000 +#define AT91C_MAIN_CLOCK 179712000 /*
- peripheral clock
- (AT91C_MASTER_CLOCK / 3)
*/ -#define AT91C_MASTER_CLOCK 59904000 +#define AT91C_MASTER_CLOCK 59904000
I wonder if it would be possible to do the calculations in the defines like #define AT91C_MAIN_CLOCK ((AT91C_XTAL_CLOCK/4)*39). You get the idea...
Sounds better ...
-#define AT91_SLOW_CLOCK 32768 /* slow clock */ +#define CONFIG_ARM920T 1 /* This is an ARM920T Core */ +#define CONFIG_AT91RM9200 1 /* It's an Atmel AT91RM9200 SoC */ +#define CONFIG_AT91RM9200EK 1 /* on an AT91RM9200EK Board */ +#define CONFIG_CPUAT91 1 /* we use newer AT91 code */ +#define USE_920T_MMU 1
-#define CONFIG_ARM920T 1 /* This is an ARM920T Core */ -#define CONFIG_AT91RM9200 1 /* It's an Atmel AT91RM9200 SoC */ -#define CONFIG_AT91RM9200EK 1 /* on an AT91RM9200EK Board */ -#undef CONFIG_USE_IRQ /* we don't need IRQ/FIQ stuff */ -#define USE_920T_MMU 1 +#define CONFIG_CMDLINE_TAG 1 /* enable passing of ATAGs */ +#define CONFIG_SETUP_MEMORY_TAGS 1 +#define CONFIG_INITRD_TAG 1
-#define CONFIG_CMDLINE_TAG 1 /* enable passing of ATAGs */ -#define CONFIG_SETUP_MEMORY_TAGS 1 -#define CONFIG_INITRD_TAG 1
On all files we touch, we want to change "#define SOMETHING 1" into "#define SOMETHING" when the value itself is not used. Please try to fix that globally.
NP
@@ -113,10 +112,6 @@ #define CONFIG_DBGU #undef CONFIG_USART0 #undef CONFIG_USART1
Don't #undef what has never been defined.
You know this is historic. This are the three possible interfaces for USART on this board. At least at91rm9200 known about DBGU ;) This will be addressed when merging at91rm9200/at91 and maybe avr32 usart implementation.
+#include<asm/arch/hardware.h> /* needed for port definitions */
Are you sure that is needed in the midst of <board>.h?
Should move some lines up ...
Regards
Andreas Bießmann