Hello!

I'm trying to get to get the AT91SAM9G45 qualified for my use. Which involves reading an SD card in U-Boot, but am not having much luck with that.
Starting with 2009.08, I've applied the following patches:
http://lists.denx.de/pipermail/u-boot/2 ... 59456.html
http://lists.denx.de/pipermail/u-boot/2 ... 60053.html
http://lists.denx.de/pipermail/u-boot/2 ... 60243.html

Then added:

to the end of cpu/arm926ejs/at91/cpu.c based on this:
http://lists.denx.de/pipermail/u-boot/2 ... 59457.html

Then added a call to the initialize function from the board init code in:
board/atmel/at91sam9m10g45ek/at91sam9m10g45ek.c


In the board configuration file, include/configs/at91sam9m10g45ek.h, I added:


And see things show up at the start of U-Boot:

(I'm not sure about that Unknown FLASH, I originally thought it was the DataFlash, but pulling the chip select jumper didn't make it go away.)

However, executing 'mmc init' and 'mmcinfo' yield no results:


I've compared the U-Boot at91sam9m10g45_devices.c and Linux's at91sam9g45_devices.c and they seem to agree with how to configure the MCI device in the chip, and the SD card works under Linux. I'm not sure how to proceed from here.

Thanks,
Dan

Hi,

Did you already solved the problem? If you still have the problem register to the u-boot devel maillinglist and aks again.
Hopefully I have the same problem like you in about 4 days and than the can discuss it there.

bye
Konrad

Hi,

Has anyone managed to get the mmc working in U-boot for the at91sam9m10g45-ek? The hardware works fine under Linux, so it's just the U-boot driver that needs to be fixed. I tried various CONFIG_ arrangements for u-boot-2011.12 based on Web search of postings, but wasn't successful in getting a working mmc. I guess if this hasn't been done, then I'll have to try to get the U-boot mmc to work the same way as that in Linux.

Thanks for help.

Peter Lu

Turns out the MCI controller on the at91sam9m10g45-ek doesn't do multi-block transfers in PIO mode very well, freezing up when the transfer-abort CMD12 is issued. So, it's just easier to have the U-boot driver use single-block transfers. The necessary configurations applied to u-boot-2011.12 are shown below...




== In at91sam9m10g45ek.h

#define CONFIG_CMD_MMC
#define CONFIG_CMD_FAT

/* MMC/SD */
#define CONFIG_MMC 1
#define CONFIG_GENERIC_MMC 1
#define CONFIG_GENERIC_ATMEL_MCI 1
#define CONFIG_SYS_MMC_MAX_BLK_COUNT 1 // Use single transfers, as multi-transfers hang




== In at91sam9m10g45ek.c

#if defined(CONFIG_GENERIC_ATMEL_MCI)
# include <mmc.h>
#endif


...



#ifdef CONFIG_GENERIC_ATMEL_MCI
void at91_mci_hw_init_dev(int dev)
{
if (dev == 0) {
at91_set_A_periph(AT91_PIN_PA0, 0); /* MCCKA */
at91_set_A_periph(AT91_PIN_PA1, 1); /* MCCDA */
at91_set_A_periph(AT91_PIN_PA2, 1); /* MCDA0 */
at91_set_A_periph(AT91_PIN_PA3, 1); /* MCDA1 */
at91_set_A_periph(AT91_PIN_PA4, 1); /* MCDA2 */
at91_set_A_periph(AT91_PIN_PA5, 1); /* MCDA3 */
}
else {
at91_set_A_periph(AT91_PIN_PA31, 0); /* MCCKB */
at91_set_A_periph(AT91_PIN_PA22, 1); /* MCCDB */
at91_set_A_periph(AT91_PIN_PA23, 1); /* MCDB0 */
at91_set_A_periph(AT91_PIN_PA24, 1); /* MCDB1 */
at91_set_A_periph(AT91_PIN_PA25, 1); /* MCDB2 */
at91_set_A_periph(AT91_PIN_PA26, 1); /* MCDB3 */
}
}

int board_mmc_init(bd_t *bd)
{
struct at91_pmc *pmc = (struct at91_pmc *)ATMEL_BASE_PMC;

writel(1 << ATMEL_ID_MCI0, &pmc->pcer); // enable MCI clock
at91_mci_hw_init_dev(0);
atmel_mci_init((void *)ATMEL_BASE_MCI0); // in gen_atmel_mci.c

writel(1 << ATMEL_ID_MCI1, &pmc->pcer); // enable MCI clock
at91_mci_hw_init_dev(1);
atmel_mci_init((void *)ATMEL_BASE_MCI1); // in gen_atmel_mci.c

return 0;
}
#endif


== Suggested but not necessary changes in gen_atmel_mci.c for cleanliness

81c81
< debug("mci: setting clock %u Hz, block size %u\n",
---
> printf("mci: setting clock %u Hz, block size %u\n",
243c243
< print_buffer(0, data->dest + block_count * data->blocksize, 1,
---
> print_buffer(0, data->dest, 1,