Goodday Magnus!

In message on Saturday 03 January 2009 03:49:53 Magnus Lilja wrote:

Hi

I have used your patch to boot the i.MX31 PDK board from its large page NAND now. I had to modify one function (see below).

2008/12/5 Maxim Artamonov scn1874@yandex.ru:

...

V2 NAND_SPL support for phycore imx31

Changelog:

• Added bad block verify and skip;
• Correct code to codestyle;
• Added few comments;

diff --git a/nand_spl/nand_boot_mx31.c b/nand_spl/nand_boot_mx31.c +static void mx31_nand_data_output(void) +{

•   NFC_CONFIG1 = NFC_ECC_EN;
  

•   NFC_BUF_ADDR = 0; /* read in first 0 buffer*/
  

•   NFC_CONFIG2 = NFC_OUTPUT;
  

•   mx31_wait_ready();
  

+}

I suggest using the following function instead in order to support for small page and large page devices:

static void mx31_nand_data_output(void) { int i;

/* * The NAND controller requires four output commands for * large page devices. */ for (i = 0; i < (CFG_NAND_PAGE_SIZE / 512); i++) { NFC_CONFIG1 = NFC_ECC_EN; NFC_BUF_ADDR = i; /* read in i:th buffer */ NFC_CONFIG2 = NFC_OUTPUT; mx31_wait_ready(); } }

Ok, I have'nt possibilities for check this, but I think you have checked this. I'll add you code.

...

•   mx31_nand_data_output ();/*fill the main buffer 0*/
  

•   if (mx31_nand_check_ecc ())
  

•           while (1){
  

•                   break;};        /*!!!!!*/
  

How about returning an error code and letting the nand_boot() function below just loop forever (or something) if this happens?

I'll add infinitive loop for this. Return code of error to ..?

...

•   p1 = (u32 *)MAIN_AREA0;
  

•   p2 = (u32 *)buf;
  

•   /*main copy loop from NAND-buffer to SDRAM memory*/
  

•   for (i=0; i < (CFG_NAND_PAGE_SIZE / 4); i++){
  

•           *p2 = *p1;
  

•           p1++;
  

•           p2++;
  

•   }
  

•   p1 = (u32 *)SPARE_AREA0;
  

•   /*it is hardware specific code for 8-bit 512kB NAND-flash spare area*/
  

I guess "512kB" should be just "512"?

You are right. 512 B.

...

•   p1++;
  

•   a = *p1;
  

•   a = (a & 0x0000ff00) >> 8;
  

Is this code valid also for large page NAND devices? I don't know the answer myself.

No, for 2-k paged Nand it would be better to write something like that:

a = *p1; a = (a & 0x000000ff);

In addition, there is a difference between 8-bit and 16-bit modes.

I hope you have the possibility to update the patch with the above as well as Scott's comments.

I'll do.

with regards, Maxim