11 Dec
2008
11 Dec
'08
8:23 p.m.
On Thu, Dec 11, 2008 at 07:47:18AM -0500, Mike Frysinger wrote:
btw, does nand_init() fall into the same category ?
Hmm, in general I'd say yes, but that raises the issue with certain drivers (such as fsl elbc) of how to communicate the ECC layout being used to the kernel. Currently (or at least, with a patch that was posted) the kernel will leave u-boot's configuration alone. Perhaps we should add a device tree property.
should i convert that to a "nand init" step ?
No, if initialization is deferred have it happen automatically when the first real nand command is run. We already defer the bad block scan in this manner.
-Scott