On Tuesday, October 26, 2010 3:50 AM, David Woodhouse wrote: > On Tue, 2010-10-26 at 09:51 +0100, David Woodhouse wrote: >> On Tue, 2010-10-26 at 13:49 +1100, Stephen Rothwell wrote: >>> >>> I am not sure where these came from ... >> >> It's a config issue. You have chosen to build NOR flash support. >> >> NOR flash can be wired up many ways. It could be four 8-bit chips in >> parallel in a 32-bit bus, two 16-bit chips in a 32-bit bus, 2 8-bit >> chips in a 16-bit bus, one 16-bit chip in a 16-bus bus, etc... >> >> You have elected to support *none* of those configurations. No NOR chip >> support can work. >> >> You don't get to choose individual geometry options unless you set >> CONFIG_MTD_ADV_OPTIONS -- by default you get a fully versatile driver >> which can cope with anything at runtime. >> >> I'll build a ppc44x_defconfig and work out what's going wrong. I'd >> expect this kind of thing with randconfig, but not a defconfig. > > Fixed by reverting the offending commit: > http://git.infradead.org/mtd-2.6.git/commitdiff/6411bf6cd I didn't follow the logic for the MTD_MAP_BANK_WIDTH_* all the way thru. Honestly I still don't quite get it. If MTD_CFI_GEOMETRY is n wouldn't all the MTD_MAP_BANK_WIDTH_* also be n? Sorry about the trouble. Hartleyÿô.nlj·Ÿ®‰†+%ŠË±é¥Šwÿº{.nlj·¥Š{±þw±·ø¡Ü}©ž²ÆzÚj:+v‰¨þø®w¥þŠàÞ¨è&¢)ß«a¶Úÿûz¹ÞúŽŠÝjÿŠwèf
