08.02.2019 13:10, Kejia Hu пишет: > On Thu, 2019-02-07 at 16:23 +0300, Dmitry Osipenko wrote: >> this all looks very generic and much more board-specific than automotive > > Hi, > > Thanks for your reply. > > We got the clock data from nvidia pdk for the automotive which didn't > set the clocks up in the same way as commercial parts, and we were told > that there are certain clock restrictions for the automotive parts that > the commercial ones do not have. > > Applying different clock configs for each nodes doesn't seems to be a > good way, as they will be too scattered in the device tree. We proposed > to add a new root node for the clock originally[1], and further > discussion suggested us to distinguish the variation by reading the SKU > info, thus came this patch. We can discuss further though. > > Regards > > [1] https://www.spinics.net/lists/arm-kernel/msg665344.html > I don't see any problems with specifying clock configs for each node in the device tree. It also looks to me that the clock configs you're suggesting in this series are only relevant for a one specific board (or maybe family). You could move out the assigned-clock configs into a separate DTSI file that will be shared by all of "automotive" boards. Maybe it will make sense to have a custom init-table to avoid some of the fatal clock configurations during of early boot of the kernel. But maybe in this case we could just limit those vital clocks to a saner-common value?
