Revert "dt-bindings: pinctrl: bcm4708-pinmux: rework binding to use syscon"
Subsystems (5)
PATCH v2 1/2 Revert "dt-bindings: pinctrl: bcm4708-pinmux: rework binding to use syscon"
From: Rafał Miłecki <[email protected]>
This reverts commit 2ae80900f239484069569380e1fc4340fd6e0089.
My rework was unneeded & wrong. It replaced a clear & correct "reg" property usage with a custom "offset" one.
Back then I didn't understand how to properly handle CRU block binding. I heard / read about syscon and tried to use it in a totally invalid way. That change also missed Rob's review (obviously).
Northstar's pin controller is a simple consistent hardware block that can be cleanly mapped using a 0x24 long reg space.
Since the rework commit there wasn't any follow up modifying in-kernel DTS files to use the new binding. Broadcom also isn't known to use that bugged binding. There is close to zero chance this revert may actually cause problems / regressions.
This commit is a simple revert. Example binding may (should) be updated / cleaned up but that can be handled separately.
---
V2: Update brcm,cru.yaml to avoid new yamllint warnings/errors
--- .../devicetree/bindings/mfd/brcm,cru.yaml | 11 +++++---- .../bindings/pinctrl/brcm,ns-pinmux.yaml | 23 +++++++++++-------- 2 files changed, 19 insertions(+), 15 deletions(-)
Documentation/devicetree/bindings/mfd/brcm,cru.yaml +6 -5
@@ -32,13 +32,13 @@ properties: "#size-cells": const: 1 - pinctrl: - $ref: ../pinctrl/brcm,ns-pinmux.yaml - patternProperties: '^clock-controller@[a-f0-9]+$': $ref: ../clock/brcm,iproc-clocks.yaml + '^pin-controller@[a-f0-9]+$': + $ref: ../pinctrl/brcm,ns-pinmux.yaml + '^thermal@[a-f0-9]+$': $ref: ../thermal/brcm,ns-thermal.yaml
@@ -73,9 +73,10 @@ examples: "iprocfast", "sata1", "sata2"; }; - pinctrl { + pin-controller@1c0 { compatible = "brcm,bcm4708-pinmux"; - offset = <0x1c0>; + reg = <0x1c0 0x24>; + reg-names = "cru_gpio_control"; }; thermal@2c0 {
Documentation/devicetree/bindings/pinctrl/brcm,ns-pinmux.yaml +13 -10
@@ -17,9 +17,6 @@ description: A list of pins varies across chipsets so few bindings are available. - Node of the pinmux must be nested in the CRU (Central Resource Unit) "syscon" - node. - properties: compatible: enum:
@@ -27,10 +24,11 @@ properties: - brcm,bcm4709-pinmux - brcm,bcm53012-pinmux - offset: - description: offset of pin registers in the CRU block + reg: maxItems: 1 - $ref: /schemas/types.yaml#/definitions/uint32-array + + reg-names: + const: cru_gpio_control patternProperties: '-pins$':
@@ -72,19 +70,24 @@ allOf: uart1_grp ] required: - - offset + - reg + - reg-names additionalProperties: false examples: - | cru@1800c100 { - compatible = "syscon", "simple-mfd"; + compatible = "simple-bus"; reg = <0x1800c100 0x1a4>; + ranges; + #address-cells = <1>; + #size-cells = <1>; - pinctrl { + pin-controller@1c0 { compatible = "brcm,bcm4708-pinmux"; - offset = <0xc0>; + reg = <0x1c0 0x24>; + reg-names = "cru_gpio_control"; spi-pins { function = "spi";
Signature · 1 line
2.26.2
PATCH v2 2/2 Revert "pinctrl: bcm: ns: support updated DT binding as syscon subnode"
From: Rafał Miłecki <[email protected]>
This reverts commit a49d784d5a8272d0f63c448fe8dc69e589db006e.
The updated binding was wrong / invalid and has been reverted. There isn't any upstream kernel DTS using it and Broadcom isn't known to use it neither. There is close to zero chance this will cause regression for anyone.
Actually in-kernel bcm5301x.dtsi still uses the old good binding and so it's broken since the driver update. This revert fixes it.
--- drivers/pinctrl/bcm/pinctrl-ns.c | 29 ++++++++++------------------- 1 file changed, 10 insertions(+), 19 deletions(-)
drivers/pinctrl/bcm/pinctrl-ns.c +10 -19
@@ -5,7 +5,6 @@#include <linux/err.h> #include <linux/io.h> -#include <linux/mfd/syscon.h> #include <linux/module.h> #include <linux/of.h> #include <linux/of_device.h>
@@ -13,7 +12,6 @@#include <linux/pinctrl/pinctrl.h> #include <linux/pinctrl/pinmux.h> #include <linux/platform_device.h> -#include <linux/regmap.h> #include <linux/slab.h> #define FLAG_BCM4708 BIT(1)
@@ -24,8 +22,7 @@ struct ns_pinctrl {struct device *dev;unsigned int chipset_flag;struct pinctrl_dev *pctldev; -struct regmap *regmap; -u32 offset; +void __iomem *base;struct pinctrl_desc pctldesc;struct ns_pinctrl_group *groups;
@@ -232,9 +229,9 @@ static int ns_pinctrl_set_mux(struct pinctrl_dev *pctrl_dev, unset |=BIT (pin_number); } -regmap_read (ns_pinctrl->regmap, ns_pinctrl->offset, &tmp); + tmp =readl (ns_pinctrl->base); tmp &= ~unset; -regmap_write (ns_pinctrl->regmap, ns_pinctrl->offset, tmp); +writel (tmp, ns_pinctrl->base);return 0 ; }
@@ -266,13 +263,13 @@ static const struct of_device_id ns_pinctrl_of_match_table[] = {static int ns_pinctrl_probe (struct platform_device *pdev) {struct device *dev = &pdev->dev; -struct device_node *np = dev->of_node;const struct of_device_id *of_id;struct ns_pinctrl *ns_pinctrl;struct pinctrl_desc *pctldesc;struct pinctrl_pin_desc *pin;struct ns_pinctrl_group *group;struct ns_pinctrl_function *function; +struct resource *res;int i; ns_pinctrl =devm_kzalloc (dev,sizeof (*ns_pinctrl),GFP_KERNEL );
@@ -290,18 +287,12 @@ static int ns_pinctrl_probe(struct platform_device *pdev)return -EINVAL ; ns_pinctrl->chipset_flag = (uintptr_t )of_id->data; - ns_pinctrl->regmap =syscon_node_to_regmap (of_get_parent (np)); -if (IS_ERR (ns_pinctrl->regmap)) { -int err =PTR_ERR (ns_pinctrl->regmap); - -dev_err (dev,"Failed to map pinctrl regs: %d\n" , err); - -return err; - } - -if (of_property_read_u32 (np,"offset" , &ns_pinctrl->offset)) { -dev_err (dev,"Failed to get register offset\n" ); -return -ENOENT ; + res =platform_get_resource_byname (pdev,IORESOURCE_MEM , +"cru_gpio_control" ); + ns_pinctrl->base =devm_ioremap_resource (dev, res); +if (IS_ERR (ns_pinctrl->base)) { +dev_err (dev,"Failed to map pinctrl regs\n" ); +return PTR_ERR (ns_pinctrl->base); }memcpy (pctldesc, &ns_pinctrl_desc,sizeof (*pctldesc));
Signature · 1 line
2.26.2
Rob Herring
maintainerQuoted 22 lines · Rafał Miłecki
From: Rafał Miłecki <[email protected]>
This reverts commit
2ae80900f239484069569380e1fc4340fd6e0089.My rework was unneeded & wrong. It replaced a clear & correct "reg" property usage with a custom "offset" one.
Back then I didn't understand how to properly handle CRU block binding. I heard / read about syscon and tried to use it in a totally invalid way. That change also missed Rob's review (obviously).
Northstar's pin controller is a simple consistent hardware block that can be cleanly mapped using a
0x24long reg space.Since the rework commit there wasn't any follow up modifying in-kernel DTS files to use the new binding. Broadcom also isn't known to use that bugged binding. There is close to zero chance this revert may actually cause problems / regressions.
This commit is a simple revert. Example binding may (should) be updated / cleaned up but that can be handled separately.
Quoted 8 lines
Signed-off-by Rafał Miłecki <[email protected]>---
V2: Update brcm,cru.yaml to avoid new yamllint warnings/errors--- .../devicetree/bindings/mfd/brcm,cru.yaml | 11 +++++---- .../bindings/pinctrl/brcm,ns-pinmux.yaml | 23 +++++++++++-------- 2 files changed, 19 insertions(+), 15 deletions(-)
My bot found errors running 'make DT_CHECKER_FLAGS=-m dt_binding_check' on your patch (DT_CHECKER_FLAGS is new in v5.13):
yamllint warnings/errors:
dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/pinctrl/brcm,ns-pinmux.example.dt.yaml: cru@1800c100: $nodename:0: 'cru@1800c100' does not match '^([a-z][a-z0-9\\-]+-bus|bus|soc|axi|ahb|apb)(@[0-9a-f]+)?$'
From schema: /usr/local/lib/python3.8/dist-packages/dtschema/schemas/simple-bus.yaml
doc reference errors (make refcheckdocs):
See https://patchwork.ozlabs.org/patch/1538413
This check can fail if there are any dependencies. The base for a patch series is generally the most recent rc1.
If you already ran 'make dt_binding_check' and didn't see the above error(s), then make sure 'yamllint' is installed and dt-schema is up to date:
pip3 install dtschema --upgrade
Please check and re-submit.
Quoted 32 lines · Rob Herring
Quoted 22 lines · Rafał Miłecki
From: Rafał Miłecki <[email protected]>
This reverts commit
2ae80900f239484069569380e1fc4340fd6e0089.My rework was unneeded & wrong. It replaced a clear & correct "reg" property usage with a custom "offset" one.
Back then I didn't understand how to properly handle CRU block binding. I heard / read about syscon and tried to use it in a totally invalid way. That change also missed Rob's review (obviously).
Northstar's pin controller is a simple consistent hardware block that can be cleanly mapped using a
0x24long reg space.Since the rework commit there wasn't any follow up modifying in-kernel DTS files to use the new binding. Broadcom also isn't known to use that bugged binding. There is close to zero chance this revert may actually cause problems / regressions.
This commit is a simple revert. Example binding may (should) be updated / cleaned up but that can be handled separately.
Quoted 8 lines
Signed-off-by Rafał Miłecki <[email protected]>---
V2: Update brcm,cru.yaml to avoid new yamllint warnings/errors--- .../devicetree/bindings/mfd/brcm,cru.yaml | 11 +++++---- .../bindings/pinctrl/brcm,ns-pinmux.yaml | 23 +++++++++++-------- 2 files changed, 19 insertions(+), 15 deletions(-)
Quoted 8 lines
My bot found errors running 'make DT_CHECKER_FLAGS=-m
dt_binding_check' on your patch (DT_CHECKER_FLAGSis new in v5.13):yamllint warnings/errors:
dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/pinctrl/brcm,ns-pinmux.example.dt.yaml: cru@1800c100: $nodename:0: 'cru@1800c100' does not match '^([a-z][a-z0-9\\-]+-bus|bus|soc|axi|ahb|apb)(@[0-9a-f]+)?$'
From schema:/usr/local/lib/python3.8/dist-packages/dtschema/schemas/simple-bus.yaml
It's warning we already have and not something introduced by this revert.
As a revert this commit should introduce as little non-revert changes as possible. I'm planning to improve that example later in a separated commit.
Can you take a look at this commit despite your bot warning, please?
Rob Herring
maintainerQuoted 36 lines · Rafał Miłecki
Quoted 35 lines · Rob Herring
Quoted 22 lines · Rafał Miłecki
From: Rafał Miłecki <[email protected]>
This reverts commit
2ae80900f239484069569380e1fc4340fd6e0089.My rework was unneeded & wrong. It replaced a clear & correct "reg" property usage with a custom "offset" one.
Back then I didn't understand how to properly handle CRU block binding. I heard / read about syscon and tried to use it in a totally invalid way. That change also missed Rob's review (obviously).
Northstar's pin controller is a simple consistent hardware block that can be cleanly mapped using a
0x24long reg space.Since the rework commit there wasn't any follow up modifying in-kernel DTS files to use the new binding. Broadcom also isn't known to use that bugged binding. There is close to zero chance this revert may actually cause problems / regressions.
This commit is a simple revert. Example binding may (should) be updated / cleaned up but that can be handled separately.
Quoted 8 lines
Signed-off-by Rafał Miłecki <[email protected]>---
V2: Update brcm,cru.yaml to avoid new yamllint warnings/errors--- .../devicetree/bindings/mfd/brcm,cru.yaml | 11 +++++---- .../bindings/pinctrl/brcm,ns-pinmux.yaml | 23 +++++++++++-------- 2 files changed, 19 insertions(+), 15 deletions(-)My bot found errors running 'make DT_CHECKER_FLAGS=-m
dt_binding_check' on your patch (DT_CHECKER_FLAGSis new in v5.13):
Quoted 8 lines
Quoted 5 lines
yamllint warnings/errors:
dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/pinctrl/brcm,ns-pinmux.example.dt.yaml: cru@1800c100: $nodename:0: 'cru@1800c100' does not match '^([a-z][a-z0-9\\-]+-bus|bus|soc|axi|ahb|apb)(@[0-9a-f]+)?$'
From schema:/usr/local/lib/python3.8/dist-packages/dtschema/schemas/simple-bus.yamlIt's warning we already have and not something introduced by this revert.
We don't allow warnings, so you're saying this one was warning before?
Quoted 3 lines
As a revert this commit should introduce as little non-revert changes as possible. I'm planning to improve that example later in a separated commit.
You have to fix the warning as part of this series.
Quoted 2 lines
Can you take a look at this commit despite your bot warning, please?
If 'CRU' is a hardware block, then I don't think most of the revert is right to use 'simple-bus'.
If you want to just replace 'offset' with 'reg', then I'd be happy to see that. I've been pushing child nodes of syscons in that direction.
Rob
Quoted 39 lines · Rob Herring
Quoted 38 lines · Rafał Miłecki
Quoted 37 lines · Rob Herring
Quoted 22 lines · Rafał Miłecki
From: Rafał Miłecki <[email protected]>
This reverts commit
2ae80900f239484069569380e1fc4340fd6e0089.My rework was unneeded & wrong. It replaced a clear & correct "reg" property usage with a custom "offset" one.
Back then I didn't understand how to properly handle CRU block binding. I heard / read about syscon and tried to use it in a totally invalid way. That change also missed Rob's review (obviously).
Northstar's pin controller is a simple consistent hardware block that can be cleanly mapped using a
0x24long reg space.Since the rework commit there wasn't any follow up modifying in-kernel DTS files to use the new binding. Broadcom also isn't known to use that bugged binding. There is close to zero chance this revert may actually cause problems / regressions.
This commit is a simple revert. Example binding may (should) be updated / cleaned up but that can be handled separately.
Quoted 8 lines
Signed-off-by Rafał Miłecki <[email protected]>---
V2: Update brcm,cru.yaml to avoid new yamllint warnings/errors--- .../devicetree/bindings/mfd/brcm,cru.yaml | 11 +++++---- .../bindings/pinctrl/brcm,ns-pinmux.yaml | 23 +++++++++++-------- 2 files changed, 19 insertions(+), 15 deletions(-)My bot found errors running 'make DT_CHECKER_FLAGS=-m
dt_binding_check' on your patch (DT_CHECKER_FLAGSis new in v5.13):yamllint warnings/errors:
Quoted 8 lines
Quoted 6 lines
Quoted 3 lines
dtschema/dtc warnings/errors:
/builds/robherring/linux-dt-review/Documentation/devicetree/bindings/pinctrl/brcm,ns-pinmux.example.dt.yaml: cru@1800c100: $nodename:0: 'cru@1800c100' does not match '^([a-z][a-z0-9\\-]+-bus|bus|soc|axi|ahb|apb)(@[0-9a-f]+)?$'
From schema:/usr/local/lib/python3.8/dist-packages/dtschema/schemas/simple-bus.yamlIt's warning we already have and not something introduced by this revert.
We don't allow warnings, so you're saying this one was warning before?
Yes.
Quoted 5 lines
Quoted 3 lines
As a revert this commit should introduce as little non-revert changes as possible. I'm planning to improve that example later in a separated commit.
You have to fix the warning as part of this series.
OK
Quoted 7 lines
Quoted 1 line
Can you take a look at this commit despite your bot warning, please?
If 'CRU' is a hardware block, then I don't think most of the revert is right to use 'simple-bus'.
If you want to just replace 'offset' with 'reg', then I'd be happy to see that. I've been pushing child nodes of syscons in that direction.
Well, it's a revert, not a fix, it goes from totally wrong to semi-wrong ;) Let me send V3.