On Tue, Oct 28, 2008 at 11:45:21AM +0100, Ilko Iliev wrote: > > AFAICT, it does a read using MTD_OOB_AUTO, which can span multiple > > free segments. > Yes, but the current U-BOOT uses MTD_OOB_PLACE and the command "nand > erase clean" marks all blocks as bad.
I'm not defending the current code. > >>> What if oobsize > 64 (as with 4k pages)? Why write anything at all if > >>> you're not going to write the cleanmarker? Why badblockpos & ~1 (I > >>> know > >>> existing code does it, but why)? > >>> > >> The current Linux NAND Flash driver supports 8, 16 and 64 bytes OOB. > > > > No need to add a place that will silently break if that changes, though. > No, I think it will work also with OOB=128 bytes Your patch as-is will write random data to the second half of a 128-byte OOB if it doesn't have enough space in the first free segment for the clean marker. > >>> > >> I think the NAND driver should work not only with MTD_OOB_AUTO. > > > > Explain? MTD_OOB_AUTO is a feature of the NAND subsystem, which > > automatically places user OOB data in the free regions described by > > the low-level driver. It's not some hardware feature that may or may > > not be present. > At the moment U-BOOT doesn't use MTD_OOB_AUTO and the NAND flash can't > be erased with "nand erase clean". > I think this bug should be corrected instead of to switch to MTD_OOB_AUTO. I think the bug should be corrected *by* switching to MTD_OOB_AUTO. What do you have against it? -Scott _______________________________________________ U-Boot mailing list U-Boot@lists.denx.de http://lists.denx.de/mailman/listinfo/u-boot
