On Wed, Jun 06, 2018 at 03:36:59PM -0600, Simon Glass wrote: > > This adds the sequential-write interface which provides a way to create a > device tree from scratch. It also includes various other improvements and > additions to the existing functionality. > > This requires the following v4 patches: > > pylibfdt: Return string instead of bytearray from getprop() > pylibfdt: Allow reading integer values from properties > > Note: I'm sending this now rather than waiting on comments for patch 10/10 > since there is enough rework here to be getting on with. I've now applied most of these. A few I had further comments on. One overall comment that applies to both the sw stuff and your existing rw code. At the moment, you mirror the C interface by requiring explicit resizing of the buffers holding the dt. That's a somewhat awkward interface in C, but we have a reason for it (allowing embedding in limited environments without an allocator). In Python, that interface is even odder, and there's not much reason for it - Python already requires an allocator. So, I'm wondering if it makes sense to change that interface - remove most of the 'size' parameters (except possibly as hints), trap FDT_ERR_NOSPACE errors from all the write functions and resize and retry. I've been thinking of adding an (optional) C interface that does that for ages, though I've never had time to implement it. In python it should be easier to do, though. -- David Gibson | I'll have my music baroque, and my code david AT gibson.dropbear.id.au | minimalist, thank you. NOT _the_ _other_ | _way_ _around_! http://www.ozlabs.org/~dgibson
Attachment:
signature.asc
Description: PGP signature
