It may be unreferenced, but it behaves as expected. It's part of a set
of three mutually exclusive choices, so after checking for the two
others, there's no point checking for the last one.
Of course, we could apply the patch below ;-)
Haavard
diff --git a/drivers/mtd/nand/atmel_nand.c b/drivers/mtd/nand/atmel_nand.c
index 3387e0d..d5775a9 100644
--- a/drivers/mtd/nand/atmel_nand.c
+++ b/drivers/mtd/nand/atmel_nand.c
@@ -47,6 +47,10 @@
#define no_ecc 0
#endif
+#ifdef CONFIG_MTD_NAND_ATMEL_ECC_SOFT
+/* Avoid incessant nagging */
+#endif
+
/* Register access macros */
#define ecc_readl(add, reg) \
__raw_readl(add + ATMEL_ECC_##reg)
--