Devicetree Compiler
 help / color / mirror / Atom feed
From: David Gibson <david@gibson.dropbear.id.au>
To: Simon Glass <sjg@chromium.org>
Cc: Devicetree Compiler <devicetree-compiler@vger.kernel.org>,
	Herve Codina <herve.codina@bootlin.com>
Subject: Re: [PATCH v3 09/18] dtc: Move path-reference handling out of the grammar
Date: Tue, 17 Nov 2026 10:55:38 +1100	[thread overview]
Message-ID: <avuX9aUcIXxMcn43@gractus.seuss> (raw)
In-Reply-To: <CAFLszThKSAWmf-DWaHztT9aCK6yNrHgObcusRW9XuJOZ_8QKKQ@mail.gmail.com>

[-- Attachment #1: Type: text/plain, Size: 6547 bytes --]

On Mon, Nov 16, 2026 at 04:49:42PM -0700, Simon Glass wrote:
> Hi David,
> 
> On Thu, 12 Nov 2026 at 21:09, David Gibson <david@gibson.dropbear.id.au> wrote:
> >
> > On Fri, Nov 06, 2026 at 12:51:36PM -0700, Simon Glass wrote:
> > > The grammar action for a top-level &{/path} { ... } holds its merge
> > > logic, including making an overlay fragment in a plugin.
> > >
> > > Move it into merge_ref(), which can tell a path from a label by looking
> > > at the reference, and use a single rule for both kinds of reference.
> > >
> > > The output does not change.
> > >
> > > Assisted-by: Claude:claude-opus-5-5
> > > Signed-off-by: Simon Glass <sjg@chromium.org>
> > > ---
> > >
> > > (no changes since v2)
> > >
> > > Changes in v2:
> > > - Combine merge_path_ref() into merge_ref()
> > > - Combine the rules for label and path references, using dt_ref
> > >
> > >  dtc-parser.y | 27 +--------------------------
> > >  livetree.c   | 33 +++++++++++++++++++++++----------
> > >  2 files changed, 24 insertions(+), 36 deletions(-)
> > >
> > > diff --git a/dtc-parser.y b/dtc-parser.y
> > > index 3614d71..3ec9254 100644
> > > --- a/dtc-parser.y
> > > +++ b/dtc-parser.y
> > > @@ -191,32 +191,7 @@ devicetree:
> > >                               ERROR(&@3, "Label or path %s not found", $3);
> > >                       $$ = $1;
> > >               }
> > > -     | devicetree DT_PATH_REF nodedef
> > > -             {
> > > -                     struct node *target = get_node_by_ref($1, $2);
> > > -
> > > -                     /*
> > > -                      * We rely on the rule being always:
> > > -                      *   versioninfo plugindecl memreserves devicetree
> > > -                      * so $-1 is what we want (plugindecl)
> > > -                      */
> > > -                     if ($<flags>-1 & DTSF_PLUGIN) {
> > > -                             if (is_ref_relative($2))
> > > -                                     ERROR(&@2, "Label-relative reference %s not supported in plugin", $2);
> > > -                             /* A path is in the base tree and a label may be */
> > > -                             if ($2[0] == '/' || !target)
> > > -                                     add_orphan_node($1, $3, $2);
> > > -                             else
> > > -                                     merge_nodes(target, $3);
> > > -                     } else {
> > > -                             if (target)
> > > -                                     merge_nodes(target, $3);
> > > -                             else
> > > -                                     ERROR(&@2, "Label or path %s not found", $2);
> > > -                     }
> > > -                     $$ = $1;
> > > -             }
> > > -     | devicetree DT_LABEL_REF nodedef
> > > +     | devicetree dt_ref nodedef
> > >               {
> > >                       struct fragment *f;
> > >
> > > diff --git a/livetree.c b/livetree.c
> > > index 179866d..2e9a3b2 100644
> > > --- a/livetree.c
> > > +++ b/livetree.c
> > > @@ -574,29 +574,42 @@ bool is_ref_relative(const char *ref)
> > >  }
> > >
> > >  /*
> > > - * Merge a fragment which refers to a label into the node with that label. In
> > > - * a plugin, a label which is not found becomes an overlay fragment.
> > > + * Merge a fragment which refers to a label or path into the node it refers to.
> > > + *
> > > + * In a plugin, a path always refers to the base tree and a label may do so, if
> > > + * it is not found. In both cases, the fragment becomes an overlay fragment.
> > >   *
> > >   * Returns true on success, false on an error, which has been reported.
> > >   */
> > >  bool merge_ref(struct node *dt, struct fragment *f, bool plugin)
> > >  {
> > > -     struct node *target = get_node_by_ref(dt, f->ref);
> > > +     struct node *target;
> > > +     bool ok = true;
> > >
> > >       assert(f->type == FRAGMENT_NODE);
> > > -     if (target) {
> > > -             merge_nodes(target, f->node);
> > > -             return true;
> > > +     if (plugin) {
> > > +             if (is_ref_relative(f->ref)) {
> > > +                     srcpos_error(f->ref_srcpos, "Error",
> > > +                                  "Label-relative reference %s not supported in plugin",
> > > +                                  f->ref);
> >
> > AFAICT, the fact that we can't encode label-relative references in
> > .dtbo is only relevant in the case that we can't resolve it locally.
> > So I think this should go within the next if.
> 
> Just on this one, it changes changes behaviour (a local &{label/node}
> would be merged rather than rejected), so I think it is better as its
> own patch, with a test, in the next series (the second half of what
> was in v3).

Fair enough.  Mentioning that in the commit message would be helpful.

> > > +                     ok = false;
> > > +             }
> > > +             if (f->ref[0] == '/' || !get_node_by_ref(dt, f->ref)) {
> > > +                     /* Add a fragment@N node for the reference */
> > > +                     add_orphan_node(dt, f->node, f->ref);
> >
> > Pre-existing, but what will add_orphan_node() actually *do* if given a
> > ref it has no way to encode in a dtbo?
> 
> It adds target = <&label/node>, then dies when writing __fixups__ with
> 'Can't generate fixup for reference to path &{label/node}'. The early
> error gives a position instead.

Well that's certainly an improvement, yay.

> 
> >
> > > +                     return ok;
> > > +             }
> > >       }
> > > -     if (!plugin) {
> > > +
> > > +     target = get_node_by_ref(dt, f->ref);
> >
> > You can leave the get_node_by_ref() at the top, since it's used in
> > both branches.
> >
> > > +     if (!target) {
> > >               srcpos_error(f->ref_srcpos, "Error",
> > >                            "Label or path %s not found", f->ref);
> > >               return false;
> > >       }
> > > -     /* The label may be in the base tree, so add a fragment@N node for it */
> > > -     add_orphan_node(dt, f->node, f->ref);
> > > +     merge_nodes(target, f->node);
> >
> > > -     return true;
> > > +     return ok;
> > >  }
> > >
> > >  /*
> > > --
> > > 2.43.0
> > >
> > >
> 
> Regards,
> Simon
> 

-- 
David Gibson (he or they)	| 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: 833 bytes --]

  reply	other threads:[~2026-11-16 23:57 UTC|newest]

Thread overview: 42+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2026-11-06 19:51 [PATCH v3 00/18] dtc: Merge the fragments of a dts file after parsing Simon Glass
2026-11-06 19:51 ` [PATCH v3 01/18] tests: Check how the fragments of a dts file are merged Simon Glass
2026-11-13  3:20   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 02/18] srcpos: Allow srcpos_string() to accept NULL Simon Glass
2026-11-11  4:56   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 03/18] dtc: Add a struct for the top-level fragments of a dts file Simon Glass
2026-11-13  3:46   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 04/18] dtc: Move the /delete-node/ handling out of the grammar Simon Glass
2026-11-13  3:49   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 05/18] dtc: Move the /omit-if-no-ref/ " Simon Glass
2026-11-13  3:50   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 06/18] dtc: Move is_ref_relative() to livetree.c Simon Glass
2026-11-13  3:55   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 07/18] dtc: Move label-reference handling out of the grammar Simon Glass
2026-11-13  3:58   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 08/18] dtc: Merge &{label} in a plugin as with &label Simon Glass
2026-11-13  4:01   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 09/18] dtc: Move path-reference handling out of the grammar Simon Glass
2026-11-13  4:09   ` David Gibson
2026-11-16 23:49     ` Simon Glass
2026-11-16 23:55       ` David Gibson [this message]
2026-11-06 19:51 ` [PATCH v3 10/18] dtc: Move labelled-reference " Simon Glass
2026-11-13  4:13   ` David Gibson
2026-11-16 23:49     ` Simon Glass
2026-11-16 23:57       ` David Gibson
2026-11-06 19:51 ` [PATCH v3 11/18] dtc: Move first-reference " Simon Glass
2026-11-13  4:19   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 12/18] dtc: Add merge_fragment() to merge any fragment Simon Glass
2026-11-13  4:24   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 13/18] dtc: Add helpers for the errors when merging a fragment Simon Glass
2026-11-16  3:49   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 14/18] dtc: Merge the fragments of a dts file after parsing Simon Glass
2026-11-16  4:04   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 15/18] dtc: Merge the fragments of a dts file in main() Simon Glass
2026-11-16  4:08   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 16/18] dtc: Add the overlay fragments from merge_fragments() Simon Glass
2026-11-16  4:12   ` David Gibson
2026-11-19 15:35     ` Herve Codina
2026-11-06 19:51 ` [PATCH v3 17/18] tests: Check references to a generated overlay fragment Simon Glass
2026-11-16  4:15   ` David Gibson
2026-11-06 19:51 ` [PATCH v3 18/18] dtc: Add the overlay fragments after merging a plugin Simon Glass
2026-11-16  4:34   ` David Gibson

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=avuX9aUcIXxMcn43@gractus.seuss \
    --to=david@gibson.dropbear.id.au \
    --cc=devicetree-compiler@vger.kernel.org \
    --cc=herve.codina@bootlin.com \
    --cc=sjg@chromium.org \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox