On 2018/10/19 16:10, Boris Brezillon wrote:
On Fri, 19 Oct 2018 15:29:05 +0800
Liang Yang <liang.yang@xxxxxxxxxxx> wrote:
How about defining that the HW returns an array of __le64 instead and then
define the following macros which you can use after converting in the
CPU endianness
#define ECC_GET_PROTECTED_OOB_BYTE(x, y) (((x) >> (8 * (1 + y)) & GENMASK(7, 0))
#define ECC_COMPLETE BIT(31)
#define ECC_ERR_CNT(x) (((x) >> 24) & GENMASK(5, 0))
(I'm not entirely sure the field positions are correct, but I'll let you
check that).
ok. i think it should be:
#define ECC_GET_PROTECTED_OOB_BYTE(x, y) (((x) >> (8 * y) &
GENMASK(7, 0))
if x represents the u64 and y represents the index of the u64.
Absolutely.
+
+#define PER_INFO_BYTE (sizeof(struct meson_nfc_info_format))
+
+struct meson_nfc_nand_chip {
+ struct list_head node;
+ struct nand_chip nand;
+ bool is_scramble;
I think I already mentioned the NAND_NEED_SCRAMBLING flag []. Please
drop this field and test (chip->flags & NAND_NEED_SCRAMBLING) instead.
em, i use NAND_NEED_SCRAMBLING and is_scramble is set:
static int meson_nand_attach_chip(struct nand_chip *nand)
{
......
meson_chip->is_scramble =
(nand->options & NAND_NEED_SCRAMBLING) ? 1 : 0;
......
}
Why do you need to add a new field then? Just test
nand->options & NAND_NEED_SCRAMBLING directly or provide a helper
function that does that.
ok, i will fix it.
.
______________________________________________________
Linux MTD discussion mailing list
http://lists.infradead.org/mailman/listinfo/linux-mtd/
