On Wed, 5 Dec 2018 17:32:54 +0100 Boris Brezillon <boris.brezillon@xxxxxxxxxxx> wrote: > > >>> smaller than 16MB, doesn't mean it does not support 4B opcodes. We > > >>> probably won't use the 4B opcodes in that case, but still. > > >>> > > >> > > >> I agree that manufacturers have a sense of humor and this might be possible. But > > >> there's no need to use 4B opcodes in this case, so a post_sfdp fixup will help > > >> here too. > > > > > > Except there's nothing to fix in this case, we just won't use 4B > > > opcodes if we don't need to, that's all. > > > > you'll have an extra byte of address that has a tiny impact on performance on > > small requests. I see it as a fix, we should do what's best to do. anyway ... > > No, see the checks that are done in this patch: to use 4B_CODES the > flag should be set and the NOR should be larger than 16MB. The flag > does not mean "use 4B opcodes", it meas "4B opMcodes are supported". Maybe I should rename the flag SNOR_F_SUPPORTS_4B_OPCODES to make it clear. ______________________________________________________ Linux MTD discussion mailing list http://lists.infradead.org/mailman/listinfo/linux-mtd/