On Sat, Nov 03, 2012 at 05:21:46PM +0100, Ilya Yanok wrote:
Hi Tom,
On Wed, Oct 31, 2012 at 1:03 AM, Tom Rini trini@ti.com wrote:
for (i = 0; eccsteps; eccsteps--, i += eccbytes, p += eccsize) {this->ecc.hwctl(&mtd, NAND_ECC_READ); +nand_command(block,
page, data_pos, NAND_CMD_RNDOUT); + + this->read_buf(&mtd, p, eccsize); + + nand_command(block, page, oob_pos, NAND_CMD_RNDOUT); + + this->read_buf(&mtd, oob, eccbytes); + this->ecc.calculate(&mtd, p, &ecc_calc[i]); + + data_pos += eccsize; + oob_pos += eccbytes; + oob += eccbytes; +
}This is where the function differs. If we can't merge things together, I'd like to see about putting just this function into nand_spl_simple.c under CONFIG_SYS_NAND_HW_BCH8 since if I follow what's going on, and I need to play with the code to confirm I do, it's a generic change related to how much more we're reading back out
Not exactly. This change is rather GPMC-specific: we have to read data block then it's ecc code to get the syndrome. And even with GPMC in another configuration we will need another reading order... I'm not sure if we can do this in some generic way...
OK, lets just leave that part alone for now. Maybe we need to allow for some abstraction or __weak in the SPL read case, but we can revisit this a little later on once we're able to see how say BCH16 behaves too.