On 2019/6/14 21:53, Rob Herring wrote: > On Wed, Jun 12, 2019 at 12:29 PM Frank Rowand <frowand.list@xxxxxxxxx> wrote: >> >> On 6/12/19 10:00 AM, Rob Herring wrote: >>> On Wed, Jun 12, 2019 at 10:45 AM Frank Rowand <frowand.list@xxxxxxxxx> wrote: >>>> >>>> Hi Kefeng, >>>> >>>> If Rob agrees, I'd like to see one more change in this patch. >>>> >>>> Since the only caller of of_fdt_match() is of_flat_dt_match(), >>>> can you move the body of of_fdt_match() into of_flat_dt_match() >>>> and eliminate of_fdt_match()? >>> >>> That's fine as long as we think there's never any use for of_fdt_match >>> after init? Fixup of nodes in an overlay for example. >> >> We can always re-expose the functionality as of_fdt_match() in the future >> if the need arises. But Stephen's recent patch was moving in the opposite >> direction, removing of_fdt_match() from the header file and making it >> static. > > Yes, we can, but it is just churn if we think it is likely needed. > > OTOH, we probably want users to just use libfdt API directly and > should add this to libfdt if needed. > > So yes, please implement Frank's suggestion. OK，done in patch v2. > > Rob > > . >
