Re: [PATCH v2] dtc: create tool to diff device trees

[David Gibson <david-xT8FGy+AXnRB3Ne2BGzF6laj5H9X9Tb+@public.gmane.org>]

[-- Attachment #1: Type: text/plain, Size: 3045 bytes --]

On Sun, Jan 10, 2016 at 11:56:49PM -0800, Frank Rowand wrote:
> On 1/10/2016 6:37 PM, David Gibson wrote:
> > On Thu, Jan 07, 2016 at 11:03:14AM -0800, Frank Rowand wrote:
> >> From: Frank Rowand <frank.rowand-/MT0OVThwyLZJqsBc5GL+g@public.gmane.org>
> >>
> >> Create script to diff device trees.
> >>
> >> The device tree can be in any of the forms recognized by the dtc compiler:
> >>   - source
> >>   - binary blob
> >>   - file system tree (from /proc/devicetree)
> >>
> >> If the device tree is a source file, then it is pre-processed in the
> >> same way as it would be when built in the linux kernel source tree
> >> before diffing.
> >>
> >> Signed-off-by: Frank Rowand <frank.rowand-/MT0OVThwyLZJqsBc5GL+g@public.gmane.org>
> >> ---
> >>
> >> Tools to develop and debug device tree are somewhat inadequate.  This is a
> >> small step in improving the situation.
> >>
> >> Rationale for and examples of using the script are provided in slides
> >> 1 - 78 of the elce 2015 presentation "Solving Device Tree Issues",
> >> which can be found at:
> >>
> >>    http://elinux.org/images/0/04/Dt_debugging_elce_2015_151006_0421.pdf
> >>
> >> (The script was named dtdiff instead of dtx_diff in the presentation.)
> >>
> >> Changes in v2:
> >>   - Remove dt-bindings from list of includes in cpp_flags
> >>   - Remove arch_dtc_flags, which were generated from the arch specific
> >>     dts makefile
> >>   - Reformat to 8 character tabs
> >>   - compile_to_dts(): added back missing return for binary blob
> >>
> >>
> >>  scripts/dtc/dtx_diff |  343 +++++++++++++++++++++++++++++++++++++++++++++++++++
> >>  1 file changed, 343 insertions(+)
> >>
> >> Index: b/scripts/dtc/dtx_diff
> > 
> > I think this is probably the wrong directory to put this in.  Because
> > it preprocesses in the style of the kernel, this script belongs in the
> > kernel tree, not the upstream dtc tree.  However, basically everything
> > else in this directory is imported directly from upstream dtc.
> > 
> > Putting this kernel-specific file in here will probably make updates
> > to newer upstream dtc versions more complicated.
> 
> I agree with the concern (and had the same concern when I made the choice).
> 
> One thing that makes me more comfortable with the location is that the files
> in the directory are updated with scripts/dtc/update-dtc-source.sh, which
> explicitly lists which files it copies into the Linux tree.  But even with
> that, I still admit to some discomfort with the location.
> 
> Without getting too deeply into bike shedding, does anyone have a better
> location?  And keep in mind that there are several more tools coming in
> the pipeline that would likely end up in the same location.

Hm, yeah, that's a good point.  There isn't an obvious better
candidate directory.

-- 
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 #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 819 bytes --]