On Wed, Mar 01, 2023 at 04:22:31PM +0100, Miquel Raynal wrote: > Hello, > > Following Greg's opposition to merge the current nvmem layout support > proposal [1], arguing that it would eventually grow the size of the > Linux kernel and asking for some "modularization" support, here is a > proposal to turn layout drivers into regular tristate drivers. > > The first three patches are preparation changes in order to extend (and > fix) a little bit the of/device.c support. The fix does not seem to > impact most of the current users so I guess it can live with the rest of > the series in order to avoid future merge conflicts. > > The nvmem core is then extended to support the absence of layouts and > possibly lead to probe deferrals when relevant. > > Finally, the two existing layout drivers are converted into modules and > their Kconfig symbols changed to tristate. > > The base series on which these changes apply is still contained in [1], > I would prefer to keep it as it was and apply this series on top of it. > > Tests have been conducted on a Marvell Prestera switch with the mvpp2 > Ethernet driver calling for a MAC address stored in the ONIE TLV table > available through a layout driver in an EEPROM/MTD device. > > [1] https://github.com/miquelraynal/linux/tree/nvmem-next/layouts These look sane to me, thanks for making the changes. greg k-h
