From: Heiko Stuebner <[email protected]>
On i.MX, which carries a lot of pin-groups of which most are unused on
individual boards, they noticed that this plehora of nodes also results
in the runtime-lookup-performance also degrading [0].
A i.MX-specific solution defining the pingroups in the board files but
using macros to reference the pingroup-data was not well received
This patch is trying to solve this issue in a more general way, by
adding the ability to mark nodes as needing to be referenced somewhere
in the tree.
To mark a node a needing to be referenced it must be prefixed with
/delete-unreferenced/. This makes dtc check the nodes reference-status
when creating the flattened tree, dropping it if unreferenced.
For example, the i.MX6SL pingroup
/delete-uneferenced/ pinctrl_ecspi1_1: ecspi1grp-1 {
fsl,pins = <
MX6SL_PAD_ECSPI1_MISO__ECSPI1_MISO 0x100b1
MX6SL_PAD_ECSPI1_MOSI__ECSPI1_MOSI 0x100b1
MX6SL_PAD_ECSPI1_SCLK__ECSPI1_SCLK 0x100b1
>;
};
would only be included in the dtb if it got referenced somewhere
as pingroup via
node {
pinctrl-0 <&pinctrl_ecscpi1_1>;
};
[0] http://thread.gmane.org/gmane.linux.ports.arm.kernel/275912/
Signed-off-by: Heiko Stuebner <[email protected]>
---
This is just the idea I had in [1] explored a bit more. I'm definitely
not sure if this is a valid approach to the problem.
Also this is my first venture into dtc as well as flex and bison :-) .
[1] http://www.spinics.net/lists/arm-kernel/msg303731.html
scripts/dtc/checks.c | 2 ++
scripts/dtc/dtc-lexer.l | 7 +++++++
scripts/dtc/dtc-parser.y | 5 +++++
scripts/dtc/dtc.h | 4 ++++
scripts/dtc/flattree.c | 3 +++
scripts/dtc/livetree.c | 14 ++++++++++++++
6 files changed, 35 insertions(+)
diff --git a/scripts/dtc/checks.c b/scripts/dtc/checks.c
index ee96a25..747ada8 100644
--- a/scripts/dtc/checks.c
+++ b/scripts/dtc/checks.c
@@ -472,6 +472,8 @@ static void fixup_phandle_references(struct check *c, struct node *dt,
phandle = get_node_phandle(dt, refnode);
*((cell_t *)(prop->val.val + m->offset)) = cpu_to_fdt32(phandle);
+
+ reference_node(refnode);
}
}
ERROR(phandle_references, NULL, NULL, fixup_phandle_references, NULL,
diff --git a/scripts/dtc/dtc-lexer.l b/scripts/dtc/dtc-lexer.l
index 3b41bfc..3b18e97 100644
--- a/scripts/dtc/dtc-lexer.l
+++ b/scripts/dtc/dtc-lexer.l
@@ -138,6 +138,13 @@ static int pop_input_file(void);
return DT_DEL_NODE;
}
+<*>"/delete-unreferenced/" {
+ DPRINT("Keyword: /delete-unreferenced/\n");
+ DPRINT("<PROPNODENAME>\n");
+ BEGIN(PROPNODENAME);
+ return DT_DEL_UNREFERENCED;
+ }
+
<*>{LABEL}: {
DPRINT("Label: %s\n", yytext);
yylval.labelref = xstrdup(yytext);
diff --git a/scripts/dtc/dtc-parser.y b/scripts/dtc/dtc-parser.y
index f412460..ae9108b 100644
--- a/scripts/dtc/dtc-parser.y
+++ b/scripts/dtc/dtc-parser.y
@@ -64,6 +64,7 @@ static unsigned char eval_char_literal(const char *s);
%token DT_BITS
%token DT_DEL_PROP
%token DT_DEL_NODE
+%token DT_DEL_UNREFERENCED
%token <propnodename> DT_PROPNODENAME
%token <literal> DT_LITERAL
%token <literal> DT_CHAR_LITERAL
@@ -461,6 +462,10 @@ subnode:
{
$$ = name_node(build_node_delete(), $2);
}
+ | DT_DEL_UNREFERENCED subnode
+ {
+ $$ = check_node_referenced($2);
+ }
| DT_LABEL subnode
{
add_label(&$2->labels, $1);
diff --git a/scripts/dtc/dtc.h b/scripts/dtc/dtc.h
index 3e42a07..c10c440 100644
--- a/scripts/dtc/dtc.h
+++ b/scripts/dtc/dtc.h
@@ -159,6 +159,8 @@ struct node {
int addr_cells, size_cells;
struct label *labels;
+
+ int needs_reference, is_referenced;
};
#define for_each_label_withdel(l0, l) \
@@ -193,6 +195,8 @@ struct property *reverse_properties(struct property *first);
struct node *build_node(struct property *proplist, struct node *children);
struct node *build_node_delete(void);
struct node *name_node(struct node *node, char *name);
+struct node *check_node_referenced(struct node *node);
+struct node *reference_node(struct node *node);
struct node *chain_node(struct node *first, struct node *list);
struct node *merge_nodes(struct node *old_node, struct node *new_node);
diff --git a/scripts/dtc/flattree.c b/scripts/dtc/flattree.c
index 665dad7..a327592 100644
--- a/scripts/dtc/flattree.c
+++ b/scripts/dtc/flattree.c
@@ -266,6 +266,9 @@ static void flatten_tree(struct node *tree, struct emitter *emit,
if (tree->deleted)
return;
+ if (tree->needs_reference && !tree->is_referenced)
+ return;
+
emit->beginnode(etarget, tree->labels);
if (vi->flags & FTF_FULLPATH)
diff --git a/scripts/dtc/livetree.c b/scripts/dtc/livetree.c
index b61465f..98bb33d 100644
--- a/scripts/dtc/livetree.c
+++ b/scripts/dtc/livetree.c
@@ -134,6 +134,20 @@ struct node *name_node(struct node *node, char *name)
return node;
}
+struct node *check_node_referenced(struct node *node)
+{
+ node->needs_reference = 1;
+
+ return node;
+}
+
+struct node *reference_node(struct node *node)
+{
+ node->is_referenced = 1;
+
+ return node;
+}
+
struct node *merge_nodes(struct node *old_node, struct node *new_node)
{
struct property *new_prop, *old_prop;
--
1.7.10.4
Hi,
Heiko Stübner wrote:
> From: Heiko Stuebner <[email protected]>
>
> On i.MX, which carries a lot of pin-groups of which most are unused on
> individual boards, they noticed that this plehora of nodes also results
> in the runtime-lookup-performance also degrading [0].
>
> A i.MX-specific solution defining the pingroups in the board files but
> using macros to reference the pingroup-data was not well received
>
> This patch is trying to solve this issue in a more general way, by
> adding the ability to mark nodes as needing to be referenced somewhere
> in the tree.
>
> To mark a node a needing to be referenced it must be prefixed with
> /delete-unreferenced/. This makes dtc check the nodes reference-status
> when creating the flattened tree, dropping it if unreferenced.
>
> For example, the i.MX6SL pingroup
>
> /delete-uneferenced/ pinctrl_ecspi1_1: ecspi1grp-1 {
^^
s/unef/unref/
not really important here, but bad for copy/pasters
Lothar Waßmann
--
___________________________________________________________
Ka-Ro electronics GmbH | Pascalstraße 22 | D - 52076 Aachen
Phone: +49 2408 1402-0 | Fax: +49 2408 1402-10
Geschäftsführer: Matthias Kaussen
Handelsregistereintrag: Amtsgericht Aachen, HRB 4996
http://www.karo-electronics.de | [email protected]
___________________________________________________________
Hi,
On Thu, Jan 30, 2014 at 01:25:56PM +0100, Heiko St?bner wrote:
> From: Heiko Stuebner <[email protected]>
>
> On i.MX, which carries a lot of pin-groups of which most are unused on
> individual boards, they noticed that this plehora of nodes also results
> in the runtime-lookup-performance also degrading [0].
>
> A i.MX-specific solution defining the pingroups in the board files but
> using macros to reference the pingroup-data was not well received
>
> This patch is trying to solve this issue in a more general way, by
> adding the ability to mark nodes as needing to be referenced somewhere
> in the tree.
>
> To mark a node a needing to be referenced it must be prefixed with
> /delete-unreferenced/. This makes dtc check the nodes reference-status
> when creating the flattened tree, dropping it if unreferenced.
>
> For example, the i.MX6SL pingroup
>
> /delete-uneferenced/ pinctrl_ecspi1_1: ecspi1grp-1 {
> fsl,pins = <
> MX6SL_PAD_ECSPI1_MISO__ECSPI1_MISO 0x100b1
> MX6SL_PAD_ECSPI1_MOSI__ECSPI1_MOSI 0x100b1
> MX6SL_PAD_ECSPI1_SCLK__ECSPI1_SCLK 0x100b1
> >;
> };
>
> would only be included in the dtb if it got referenced somewhere
> as pingroup via
>
> node {
> pinctrl-0 <&pinctrl_ecscpi1_1>;
> };
>
> [0] http://thread.gmane.org/gmane.linux.ports.arm.kernel/275912/
>
> Signed-off-by: Heiko Stuebner <[email protected]>
> ---
> This is just the idea I had in [1] explored a bit more. I'm definitely
> not sure if this is a valid approach to the problem.
> Also this is my first venture into dtc as well as flex and bison :-) .
>
> [1] http://www.spinics.net/lists/arm-kernel/msg303731.html
This is a nice feature to have :)
However, I wonder wether it's the right way of implementing it. I'd
have another use case for this, which would be to embed a DT in a
first-stage bootloader. We have around 7kB available currently for the
DT, and this is actually less than any of our (very small already)
DTBs. So having this of feature makes complete sense to us
too. However, we also have some other tools that require most of the
nodes to be in the DTB.
So I guess turning wether we should delete a node if unreferenced into
a dtc option would make sense, since every DT user will be able to
choose, without having to duplicate the DT and make conflicting
changes.
Maxime
--
Maxime Ripard, Free Electrons
Embedded Linux, Kernel and Android engineering
http://free-electrons.com