> >> > >> >> If so, I have to add patch for mtd subsystem to always have device for > >> master > >> >> initialized regardless of kernel flag. > >> >> Only to initialize struct device, not to create full mtd node. > >> >> > >> >> Miquel - are you agree to this? > >> > >> Conceptually yes, but please mind one thing: we do not break > >> userspace. So if you want to keep the master mtd device, fine, but you > >> need to do it in a consistent way so that people not enabling the kernel > >> flag won't get a new device in their rootfs, shifting all indexes > >> upwards. > >> > >> That being said, you are probably going in the right direction by doing > >> that. > >> > >> Thanks, > >> Miquèl > > > > I've looked into this endeavour and seemed that there a need for special > > device class and careful attention in release flow to use right class. > > It will take time to do right. > > Miquel, Rodrigo, Karthik, Lucas - may the DG NVM code be merged in the > current > > form and this device be added later? > > In general, yes. But maybe you want to select > CONFIG_MTD_PARTITIONED_MASTER (IIUC your problem). > > Thanks, > Miquèl Our target is usual distribution kernel, and it never enables CONFIG_MTD_PARTITIONED_MASTER. Anyway, I've prepared patch that creates master device always and pushed a new series revision. Miquel, if you prefer to review and push the master device patch before the whole series, I can split it out. - - Thanks, Sasha
