Thanks for the input Scott.
On Fri, Jan 15, 2010 at 4:13 PM, Scott Wood <scottw...@freescale.com> wrote:
> On Tue, Jan 12, 2010 at 09:43:55PM -0700, John Rigby wrote:
>> struct fsl_nfc_regs {
>> - u32 main_area0[128]; /* @0x000 */
>> - u32 main_area1[128];
>> - u32 main_area2[128];
>> - u32 main_area3[128];
>> - u32 spare_area0[4];
>> - u32 spare_area1[4];
>> - u32 spare_area2[4];
>> - u32 spare_area3[4];
>> - u32 reserved1[64 - 16 + 64 * 5];
>> - u16 bufsiz; /* @ 0xe00 */
>> + u8 main_area[NAND_MXC_NR_BUFS][512];
>> + u8 spare_area[NAND_MXC_NR_BUFS][NAND_MXC_SPARE_BUF_SIZE];
>
> You could do u32 ...[NAND_MXC_SPARE_BUF_SIZE / 4] to avoid the casts later
> on...
ok
>
>> diff --git a/nand_spl/nand_boot_fsl_nfc.c b/nand_spl/nand_boot_fsl_nfc.c
>> index a9df2a8..02d8330 100644
>> --- a/nand_spl/nand_boot_fsl_nfc.c
>> +++ b/nand_spl/nand_boot_fsl_nfc.c
>> @@ -26,13 +26,17 @@
>>
>> #include <common.h>
>> #include <nand.h>
>> +#ifdef CONFIG_MX31
>> #include <asm-arm/arch/mx31-regs.h>
>> +#else
>> +#include <asm-arm/arch/imx-regs.h>
>> +#endif
>
> Hmm, can't this be pushed into an arch header?
>
That is what this does but only half way. I did not want to touch
mx31 stuff. I did not want my changes to creep into other subsystems.
I could add an imx-regs.h for mx31 that simply includes mx31-regs.h.
They that could be cleaned up later.
>> #include <asm/io.h>
>> #include <fsl_nfc.h>
>>
>> -static struct fsl_nfc_regs *nfc;
>> +struct fsl_nfc_regs *nfc;
>>
>> -static void nfc_wait_ready(void)
>> +void nfc_wait_ready(void)
>
> Why non-static?
Opps, left over debug, will fix.
>
>> @@ -65,12 +91,12 @@ static void nfc_nand_page_address(unsigned int
>> page_address)
>> {
>> unsigned int page_count;
>>
>> - writew(0x00, &nfc->flash_cmd);
>> + writew(0x00, &nfc->flash_add);
>> writew(NFC_ADDR, &nfc->nand_flash_config2);
>> nfc_wait_ready();
>>
>> - /* code only for 2kb flash */
>> - if (CONFIG_SYS_NAND_PAGE_SIZE == 0x800) {
>> + /* code only for large page flash */
>> + if (CONFIG_SYS_NAND_PAGE_SIZE > 512) {
>> writew(0x00, &nfc->flash_add);
>> writew(NFC_ADDR, &nfc->nand_flash_config2);
>> nfc_wait_ready();
>> @@ -88,22 +114,38 @@ static void nfc_nand_page_address(unsigned int
>> page_address)
>> page_count = page_count >> 8;
>> } while (page_count);
>> }
>> +
>> + writew(0x00, &nfc->flash_add);
>> + writew(NFC_ADDR, &nfc->nand_flash_config2);
>> + nfc_wait_ready();
>
>
No I have not tested on mx27. I don't have any hw. If someone could
test it that would be great. Also I really don't know if the syndrome
stuff is needed. This is a straight port from KARO's linux driver
(not mainline) that I know works on mx25. If someone can verify that
it works on mx27 and/or mx31 I would like to get it accepted as is
(with fixes above). After it is in I will go back and try to clean it
up some more.
This (along with later bits) looks like it changes the behavior for existing
> chips. Have you tested it on a previously-supported chip? Maybe elaborate
> on the changes in the commit message.
>
> -Scott
>
_______________________________________________
U-Boot mailing list
U-Boot@lists.denx.de
http://lists.denx.de/mailman/listinfo/u-boot
