Skip to contentSkip to conversation

ARM: Realtek RTD1195/RTD1295 SoC info

RFC Superseded Andreas Färber 10 participants 44 messages Nov 3, 2019 – Jan 3, 2020

Subsystems (5)

v1 of 2 Superseded Latest is v2 →

lore
Share & tools
lore
#0 lore raw

cover ARM: Realtek RTD1195/RTD1295 SoC info

Hello,

This series adds a soc bus driver for Realtek RTD1195 and RTD1295 SoC families.

The detection magic for RTD1295 family was mostly borrowed from downstream code and the bit meanings are entirely undocumented. In case of RTD1293 I had to invent my own detection logic, possibly flawed.
It is expected that this driver will need to be tweaked as new models emerge.

One general consideration here is that some register accesses are not well self-contained within a block so that a syscon might in theory help - but for lack of documentation we don't really have an overview of the IP blocks and their names, starts and sizes; downstream trees just hardcoded addresses.

I therefore split off the DT change to add a second/third reg entry for now, so that we could move ahead with a basic driver initially.

We have no RTD1294 DT, so it is included here mainly for illustration of the unpredictable register dependencies affecting this binding/driver.

Using reg-names might clean this up a little but would blow up the driver code as there appears to be no handy helper function provided.

Finally, I've been struggling to find an overarching name for the SoC families. Realtek.com groups them as "Digital Home Center" - not sure whether that fits? For now I use Phoenix/Kylin/etc. with DHC only as fallback, but I wonder whether those family names should rather be soc_id than family contents?

Prepared but not included here is:
* RTD1395 family, which we don't have a DT for yet,
* RTD1619 family, which we don't have a DT for yet, Chip ID to be verified,
* RTD1319 family, which we don't have a DT for yet, with TODO for its Chip ID.

Latest experimental patches at:
https://github.com/afaerber/linux/commits/rtd1295-next

Have a lot of fun!

Cheers,
Andreas

Andreas Färber (11):
dt-bindings: soc: Add Realtek RTD1195 chip info binding
soc: Add Realtek chip info driver for RTD1195 and RTD1295
arm64: dts: realtek: rtd129x: Add chip info node
ARM: dts: rtd1195: Add chip info node
dt-bindings: soc: realtek: rtd1195-chip: Extend reg property
soc: realtek: chip: Detect RTD1296
arm64: dts: realtek: rtd129x: Extend chip-info reg with CHIP_INFO1
soc: realtek: chip: Detect RTD1293
dt-bindings: soc: realtek: rtd1195-chip: Extend reg node again
soc: realtek: chip: Detect RTD1294
arm64: dts: realtek: rtd129x: Extend chip-info reg with efuse

 .../bindings/soc/realtek/realtek,rtd1195-chip.yaml |  47 +++++
 MAINTAINERS                                        |   1 +
 arch/arm/boot/dts/rtd1195.dtsi                     |   5 +
 arch/arm64/boot/dts/realtek/rtd129x.dtsi           |   7 +
 drivers/soc/Kconfig                                |   1 +
 drivers/soc/Makefile                               |   1 +
 drivers/soc/realtek/Kconfig                        |  13 ++
 drivers/soc/realtek/Makefile                       |   2 +
 drivers/soc/realtek/chip.c                         | 190 +++++++++++++++++++++
 9 files changed, 267 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml
 create mode 100644 drivers/soc/realtek/Kconfig
 create mode 100644 drivers/soc/realtek/Makefile
 create mode 100644 drivers/soc/realtek/chip.c
Signature · 1 line
2.16.4
#1 lore raw

RFC 1/11 dt-bindings: soc: Add Realtek RTD1195 chip info binding

Define a binding for RTD1195 and later SoCs' chip info registers. Add the new directory to MAINTAINERS.

---
Note: The binding gets extended compatibly later for up to three reg entries.

 .../bindings/soc/realtek/realtek,rtd1195-chip.yaml | 32 ++++++++++++++++++++++
 MAINTAINERS                                        |  1 +
 2 files changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml new file +32 -0
new file mode 100644
index 000000000000..565ad2419553
@@ -0,0 +1,32 @@
+# SPDX-License-Identifier: (GPL-2.0-or-later OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: "http://devicetree.org/schemas/soc/realtek/realtek,rtd1195-chip.yaml#"
+$schema: "http://devicetree.org/meta-schemas/core.yaml#"
+
+title: Realtek RTD1195 chip identification
+
+maintainers:
+  - Andreas Färber <[email protected]>
+
+description: |
+  The Realtek SoCs have some registers to identify the chip and revision.
+
+properties:
+  compatible:
+    const: "realtek,rtd1195-chip"
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+    chip-info@1801a200 {
+        compatible = "realtek,rtd1195-chip";
+        reg = <0x1801a200 0x8>;
+    };
+...
MAINTAINERS +1 -0
index f33adc430230..5c61cf5a44cb 100644
@@ -2188,6 +2188,7 @@ L:	[email protected] (moderated for non-subscribers)
 S:	Maintained
 F:	arch/arm64/boot/dts/realtek/
 F:	Documentation/devicetree/bindings/arm/realtek.yaml
+F:	Documentation/devicetree/bindings/soc/realtek/
 
 ARM/RENESAS ARM64 ARCHITECTURE
 M:	Geert Uytterhoeven <[email protected]>
Signature · 1 line
2.16.4
maintainer
#2 lore raw

Re: [RFC 01/11] dt-bindings: soc: Add Realtek RTD1195 chip info binding

Quoted 25 lines · Andreas Färber

Define a binding for RTD1195 and later SoCs' chip info registers. Add the new directory to MAINTAINERS.

---
Note: The binding gets extended compatibly later for up to three reg entries.

 .../bindings/soc/realtek/realtek,rtd1195-chip.yaml | 32 ++++++++++++++++++++++
 MAINTAINERS                                        |  1 +
 2 files changed, 33 insertions(+)
 create mode 100644 Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml new file +7 -0
new file mode 100644
index 000000000000..565ad2419553
@@ -0,0 +1,32 @@
+# SPDX-License-Identifier: (GPL-2.0-or-later OR BSD-2-Clause)
+%YAML 1.2
+---
+$id: "http://devicetree.org/schemas/soc/realtek/realtek,rtd1195-chip.yaml#"
+$schema: "http://devicetree.org/meta-schemas/core.yaml#"
+
+title: Realtek RTD1195 chip identification

Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml line 17

Quoted 10 lines
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml new file +10 -0
+
+maintainers:
+  - Andreas Färber <[email protected]>
+
+description: |
+  The Realtek SoCs have some registers to identify the chip and revision.
+
+properties:
+  compatible:
+    const: "realtek,rtd1195-chip"

Don't need quotes.

Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml line 24

Quoted 7 lines
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml new file +7 -0
+
+  reg:
+    maxItems: 1
+
+required:
+  - compatible
+  - reg

Add here:

additionalProperties: false

Quoted 23 lines
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml new file +8 -0
+
+examples:
+  - |
+    chip-info@1801a200 {
+        compatible = "realtek,rtd1195-chip";
+        reg = <0x1801a200 0x8>;
+    };
+...
MAINTAINERS +1 -0
index f33adc430230..5c61cf5a44cb 100644
@@ -2188,6 +2188,7 @@ L:	[email protected] (moderated for non-subscribers)
 S:	Maintained
 F:	arch/arm64/boot/dts/realtek/
 F:	Documentation/devicetree/bindings/arm/realtek.yaml
+F:	Documentation/devicetree/bindings/soc/realtek/
 
 ARM/RENESAS ARM64 ARCHITECTURE
 M:	Geert Uytterhoeven <[email protected]>
Signature · 1 line
2.16.4
#3 lore raw

RFC 2/11 soc: Add Realtek chip info driver for RTD1195 and RTD1295

In reply to Andreas Färber

Add a soc bus driver to print chip model and revision details.

Revisions from downstream drivers/soc/realtek/rtd{119x,129x}/rtk_chip.c.

---
Naming: What to call the family vs. soc_id?

 drivers/soc/Kconfig          |   1 +
 drivers/soc/Makefile         |   1 +
 drivers/soc/realtek/Kconfig  |  13 ++++
 drivers/soc/realtek/Makefile |   2 +
 drivers/soc/realtek/chip.c   | 164 +++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 181 insertions(+)
 create mode 100644 drivers/soc/realtek/Kconfig
 create mode 100644 drivers/soc/realtek/Makefile
 create mode 100644 drivers/soc/realtek/chip.c
drivers/soc/Kconfig +1 -0
index 833e04a7835c..06ae9d97321c 100644
@@ -11,6 +11,7 @@ source "drivers/soc/imx/Kconfig"
 source "drivers/soc/ixp4xx/Kconfig"
 source "drivers/soc/mediatek/Kconfig"
 source "drivers/soc/qcom/Kconfig"
+source "drivers/soc/realtek/Kconfig"
 source "drivers/soc/renesas/Kconfig"
 source "drivers/soc/rockchip/Kconfig"
 source "drivers/soc/samsung/Kconfig"
drivers/soc/Makefile +1 -0
index 2ec355003524..1d55d838a342 100644
@@ -17,6 +17,7 @@ obj-$(CONFIG_SOC_XWAY)		+= lantiq/
 obj-y				+= mediatek/
 obj-y				+= amlogic/
 obj-y				+= qcom/
+obj-y				+= realtek/
 obj-y				+= renesas/
 obj-$(CONFIG_ARCH_ROCKCHIP)	+= rockchip/
 obj-$(CONFIG_SOC_SAMSUNG)	+= samsung/
drivers/soc/realtek/Kconfig new file +13 -0
new file mode 100644
index 000000000000..be75c1889c61
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+if ARCH_REALTEK || COMPILE_TEST
+
+config REALTEK_SOC
+	tristate "Realtek chip info"
+	default ARCH_REALTEK
+	select SOC_BUS
+	help
+	  Say 'y' here to enable support for SoC info on Realtek RTD1195 and
+	  RTD1295 SoC families.
+	  If unsure, say 'n'.
+
+endif
drivers/soc/realtek/Makefile new file +2 -0
new file mode 100644
index 000000000000..49900273905b
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+obj-$(CONFIG_REALTEK_SOC) += chip.o
drivers/soc/realtek/chip.c new file +164 -0

164 lines not on this page · raw message

Signature · 1 line
2.16.4
#4 lore raw

Re: [RFC 02/11] soc: Add Realtek chip info driver for RTD1195 and RTD1295

Am 03.11.19 um 02:36 schrieb Andreas Färber:

Quoted 9 lines
 drivers/soc/Kconfig          |   1 +
 drivers/soc/Makefile         |   1 +
 drivers/soc/realtek/Kconfig  |  13 ++++
 drivers/soc/realtek/Makefile |   2 +
 drivers/soc/realtek/chip.c   | 164 +++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 181 insertions(+)
 create mode 100644 drivers/soc/realtek/Kconfig
 create mode 100644 drivers/soc/realtek/Makefile
 create mode 100644 drivers/soc/realtek/chip.c

This patch forgets to add drivers/soc/realtek/ to MAINTAINERS.

Regards,
Andreas

Signature · 9 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#5 lore raw

PATCH 1/1 base: soc: Export soc_device_to_device() helper

In reply to Andreas Färber

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

struct soc_device is local to soc.c, so it can't be inlined into the header or into driver code.

This still handles only the case that CONFIG_SOC_BUS is enabled. Same as commit da65a1589dacc7ec44ea0557a14d70a39d991f32 ("base: soc:
Provide a dummy implementation of soc_device_match()") we'd need to provide a dummy inline implementation to cope with COMPILE_TEST, too.

Cc Lee Jones <[email protected]>
Cc Bjorn Andersson <[email protected]>
Cc Geert Uytterhoeven <[email protected]>
---
 drivers/base/soc.c | 1 +
 1 file changed, 1 insertion(+)
drivers/base/soc.c +1 -0
index 4af11a423475..72848587cd51 100644
@@ -41,6 +41,7 @@ struct device *soc_device_to_device(struct soc_device *soc_dev)
 {
 	return &soc_dev->dev;
 }
+EXPORT_SYMBOL_GPL(soc_device_to_device);
 
 static umode_t soc_attribute_mode(struct kobject *kobj,
 				struct attribute *attr,
Signature · 7 lines
2.16.4


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#6 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Quoted 3 lines · Andreas Färber

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

thanks,

greg k-h

#7 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Thanks,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#8 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Quoted 8 lines · Andreas Färber

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 5 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

Quoted 8 lines
Quoted 4 lines

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

thanks,

greg k-h

#9 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 14 lines · Andreas Färber
Quoted 13 lines

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +1 -2
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);
 

But as I said, there is still in-tree code using this helper:

arch/arm/mach-ep93xx/core.c: return soc_device_to_device(soc_dev);

Returned from ep93xx_init_soc(), which is passed through by ep93xx_init_devices():

arch/arm/mach-ep93xx/adssphere.c:       ep93xx_init_devices();
arch/arm/mach-ep93xx/edb93xx.c: ep93xx_init_devices();
arch/arm/mach-ep93xx/gesbc9312.c:       ep93xx_init_devices();
arch/arm/mach-ep93xx/micro9.c:  ep93xx_init_devices();
arch/arm/mach-ep93xx/simone.c:  ep93xx_init_devices();
arch/arm/mach-ep93xx/snappercl15.c:     ep93xx_init_devices();
arch/arm/mach-ep93xx/ts72xx.c:  ep93xx_init_devices();
arch/arm/mach-ep93xx/ts72xx.c:  ep93xx_init_devices();

arch/arm/mach-ep93xx/vision_ep9307.c: ep93xx_init_devices();

Return value unused everywhere.

arch/arm/mach-imx/cpu.c: return soc_device_to_device(soc_dev);

Used as return value of imx_soc_device_init():

arch/arm/mach-imx/mach-imx6q.c: parent = imx_soc_device_init();
arch/arm/mach-imx/mach-imx6sl.c:        parent = imx_soc_device_init();
arch/arm/mach-imx/mach-imx6sx.c:        parent = imx_soc_device_init();
arch/arm/mach-imx/mach-imx6ul.c:        parent = imx_soc_device_init();

These do a NULL check and pass it to of_platform_default_populate().

arch/arm/mach-imx/mach-imx7d.c: parent = imx_soc_device_init();

This one only does a NULL check.

arch/arm/mach-imx/mach-imx7ulp.c:
of_platform_default_populate(NULL, NULL, imx_soc_device_init());

Speaks for itself.

arch/arm/mach-mxs/mach-mxs.c: parent = soc_device_to_device(soc_dev);

Passed to of_platform_default_populate().

arch/arm/mach-omap2/id.c: parent = soc_device_to_device(soc_dev);

Used for device_create_file().

arch/arm/mach-zynq/common.c: parent = soc_device_to_device(soc_dev);

Passed to of_platform_default_populate().

drivers/soc/amlogic/meson-gx-socinfo.c: dev = soc_device_to_device(soc_dev);

Used for dev_info(). CONFIG_MESON_GX_SOCINFO is bool, thus not affected.

drivers/soc/amlogic/meson-mx-socinfo.c:
dev_info(soc_device_to_device(soc_dev), "Amlogic %s %s detected\n",

Speaks for itself. CONFIG_MESON_MX_SOCINFO is bool, thus not affected.

drivers/soc/tegra/fuse/fuse-tegra.c: return soc_device_to_device(dev);

Returned from tegra_soc_device_register(). For arm64 NULL-checked only, but also used for arm in arch/arm/mach-tegra/tegra.c:tegra_dt_init(), which passes it to of_platform_default_populate().

drivers/soc/ux500/ux500-soc-id.c: parent =
soc_device_to_device(soc_dev);

Used for device_create_file().

drivers/soc/versatile/soc-integrator.c: dev = soc_device_to_device(soc_dev);

Used for device_create_file().

drivers/soc/versatile/soc-realview.c:
device_create_file(soc_device_to_device(soc_dev), &realview_manf_attr); drivers/soc/versatile/soc-realview.c:
device_create_file(soc_device_to_device(soc_dev), &realview_board_attr); drivers/soc/versatile/soc-realview.c:
device_create_file(soc_device_to_device(soc_dev), &realview_arch_attr); drivers/soc/versatile/soc-realview.c:
device_create_file(soc_device_to_device(soc_dev), &realview_build_attr);

Speaks for itself.

So, not counting my unmerged Realtek driver,
* we have two cases of struct device being used for dev_info(), which could be cleaned up with device-less pr_info(), I could post a patch,
* frequent usage in arm/mach-* for of_platform_default_populate(), this seems most difficult to replace if we neither want to cast nor expose the struct,
* some simply unused, which could be refactored to return void, and
* some for device_create_file(), which could probably be avoided with custom_attr_group.

It also raises the question of whether new arm platforms such as RTD1195 (mach-realtek) should attempt to use of_platform_default_populate() with the soc_device somehow, or if not, whether those platforms should be refactored to consistently no longer do so?

I believe in the Broken Window Theory, i.e. fixing what we can before mistakes get copied and propagate further in code; but here I don't see a perspective for getting rid of soc_device_to_device() completely to prevent new usages, nor can I test all of those platforms myself.

Has a cleanup based on custom_attr_group been attempted already and is waiting on patches to get reviewed and merged through maintainer trees, or do we need to prepare new cleanup patches here? A search for "soc_device_to_device" on LAKML Patchwork shows only this patch of mine.

Thanks,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#10 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Am 11.11.19 um 21:10 schrieb Andreas Färber:

Quoted 44 lines

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 14 lines · Andreas Färber
Quoted 13 lines

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 184

Quoted 9 lines
drivers/soc/realtek/chip.c +1 -2
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);
 

Tested and squashed in my tree.

Quoted 2 lines

But as I said, there is still in-tree code using this helper:

[snip]

Quoted 3 lines

So, not counting my unmerged Realtek driver,
* we have two cases of struct device being used for dev_info(), which could be cleaned up with device-less pr_info(), I could post a patch,

Patch sent: https://patchwork.kernel.org/patch/11237949/ (untested)

Quoted 3 lines

* frequent usage in arm/mach-* for of_platform_default_populate(), this seems most difficult to replace if we neither want to cast nor expose the struct,

One clever way might be to implement a new of_soc_default_populate() in drivers/base/soc.c that takes a soc_device instead of device, doing the conversion inside soc.c and calling of_platform_default_populate() from there. Then we could convert present users to pass around soc_device instead of device, with a perspective to make soc_device_to_device() static inside base/soc.c.

sys_soc.h does not presently #include any OF headers, so the declaration may need to go into of_platform.h and to consider CONFIG_SOC_BUS.

Will require compile-testing for each platform and ideally some kbuild bot passes to get right, so not a quick shot.

While at it, an of_soc_device_register() variant could fill soc_device_attribute::machine in common code instead of each platform duplicating to read this from the DT root node's model property.

Quoted 1 line

* some simply unused, which could be refactored to return void, and

Patch sent: https://patchwork.kernel.org/patch/11237971/ (untested)

Quoted 13 lines

* some for device_create_file(), which could probably be avoided with custom_attr_group.

It also raises the question of whether new arm platforms such as RTD1195 (mach-realtek) should attempt to use of_platform_default_populate() with the soc_device somehow, or if not, whether those platforms should be refactored to consistently no longer do so?

I believe in the Broken Window Theory, i.e. fixing what we can before mistakes get copied and propagate further in code; but here I don't see a perspective for getting rid of soc_device_to_device() completely to prevent new usages, nor can I test all of those platforms myself.

Quoted 4 lines

Has a cleanup based on custom_attr_group been attempted already and is waiting on patches to get reviewed and merged through maintainer trees, or do we need to prepare new cleanup patches here? A search for "soc_device_to_device" on LAKML Patchwork shows only this patch of mine.

Actually I don't find a single user of custom_attr_group in linux-next, which may be because the patch introducing it is from October and people are waiting on the next -rc1 before they can merge patches using it?

Regards,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#11 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

In reply to Andreas Färber
Quoted 44 lines · Andreas Färber

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 14 lines · Andreas Färber
Quoted 13 lines

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 8 lines
drivers/soc/realtek/chip.c +1 -2
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Anyway, just use the pdev pointer here, not the soc dev as that is NOT a pointer you have control over (as is evident by the fact that you need this call to try to get it.)

I'll look at your other instances later when I've had my coffee...

thanks,

greg k-h

#12 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

Quoted 48 lines · Greg Kroah-Hartman
Quoted 47 lines · Andreas Färber

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 14 lines · Andreas Färber
Quoted 13 lines

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Best regards
Uwe

Signature · 2 lines
Pengutronix e.K.                           | Uwe Kleine-König            |
Industrial Linux Solutions                 | https://www.pengutronix.de/ |
#13 lore raw

Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Am 12.11.19 um 08:29 schrieb Uwe Kleine-König:

Quoted 56 lines · Greg Kroah-Hartman
Quoted 46 lines · Andreas Färber
Quoted 45 lines

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 12 lines · Andreas Färber
Quoted 11 lines

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Quoted 6 lines

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Seconded. For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs. People who know can check /sys/bus/soc/devices/soc0, but ordinary users will at most upload dmesg output to a Bugzilla ticket.

And it was precisely info-level boot output from the Amlogic GX driver that made me aware of this common framework and inspired me to later contribute such a driver elsewhere myself. That's not a bad effect. ;)

So if anything, we should standardize that output and move it into soc_device_register(): "<family> <soc_id> [rev <revision>] detected" with suitable NULL checks? (what I did above for Realtek, minus hex)

The info level seems correct to me - it allows people to use a different log_level if they only care about warnings or errors on the console; it's not debug info for that driver, except in my case the accompanying hex numbers that I'd be happy to drop in favor of standardization.

Another aspect here is that the overall amount of soc_device_register() users is just an estimated one third above the analysis shared before. In particular server platforms, be it arm64 or x86_64, that potentially have more than one SoC integrated in a multi-socket configuration, don't feed into this soc bus at all! Therefore my comment that dev_info()-printed "soc soc0:" is kind of useless if there's only one SoC on those boards. I'm not aware of any tool or a more common file aggregating this information, certainly not /proc/cpuinfo and I'm not aware of any special "lssoc" tool either. And if there's no ACPI/DMI driver feeding support-relevant information into this framework to be generally useful, I don't expect the big distros to spend time on improving its usability.

So if we move info output into base/soc.c, we could continue using dev_info() with "soc"-grep'able prefix in the hopes that someday we'll have more than one soc device on the bus and will need to distinguish; otherwise yes, pr_info() would change the output format for Amlogic (and so would a harmonization of the text), but does anyone really care in practice? Tools shouldn't be relying on its output format, and humans will understand equally either way.

Finally, arch/arm seems unique in that it has the machine_desc mechanism that allows individual SoCs to force creating their soc_device early and using it as parent for further DT-created platform_devices. With arm64 we've lost that ability, and nios2 is not using it either.
A bad side effect (with SUSE hat on) is that this parent design pattern does not allow to build such drivers as modules, which means that distro configs using arm's multi-platform, e.g., CONFIG_ARCH_MULTI_V7 will get unnecessary code creep as new platforms get added over time (beyond the basic clk, pinctrl, tty and maybe timer).
Even if it were possible to call into a driver module that early, using it as parent would seem to imply that all the references by its children would not allow to unload the module, which I'd consider a flawed design for such an "optional" read-once driver. If we want the device hierarchy to have a soc root then most DT based platforms will have a /soc DT node anyway (although no device_type = "soc") that we probably could have a device registered for in common code rather than each SoC platform handling that differently? That might then make soc_register_device() not the creator of the device (if pre-existent) but the supplier of data to that core device, which should then allow to unload the data provider with just the sysfs data disappearing until re-inserted (speeding up the develop-and-test cycle on say not-so-resource-constrained platforms).

On the other hand, one might argue that such information should just be parsed by EBBR-conformant bootloaders and be passed to the kernel via standard UEFI interfaces and DMI tables. But I'm not aware of Barebox having implemented any of that yet, and even for U-Boot (e.g., Realtek based consumer devices...) not everyone has the GPL sources or tools to update their bootloader. So, having the kernel we control gather information relevant to kernel developers does make some sense to me.

Thoughts?

Regards,
Andreas

P.S. Sorry that a seemingly trivial one-line 0-day fix derailed into this fundamental use case discussion.

arch/arm/mach-ep93xx/core.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-imx/cpu.c:        soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-mvebu/mvebu-soc-id.c:     soc_dev =
soc_device_register(soc_dev_attr);
arch/arm/mach-mxs/mach-mxs.c:   soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-omap2/id.c:       soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-tegra/tegra.c:    struct device *parent =
tegra_soc_device_register();
arch/arm/mach-zynq/common.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/nios2/platform/platform.c:         soc_dev =

soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-gx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-mx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);

drivers/soc/atmel/soc.c:        soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/bcm/brcmstb/common.c:       soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/fsl/guts.c: soc_dev = soc_device_register(&soc_dev_attr);
drivers/soc/imx/soc-imx-scu.c:  soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/imx/soc-imx8.c:     soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/qcom/socinfo.c:     qs->soc_dev =

soc_device_register(&qs->attr);
drivers/soc/realtek/chip.c: soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/renesas/renesas-soc.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/samsung/exynos-chipid.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/tegra/fuse/fuse-tegra.c: dev = soc_device_register(attr);
drivers/soc/ux500/ux500-soc-id.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-integrator.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-realview.c: soc_dev =
soc_device_register(soc_dev_attr);

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
maintainer
#14 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Quoted 87 lines · Andreas Färber

Am 12.11.19 um 08:29 schrieb Uwe Kleine-König:

Quoted 56 lines · Greg Kroah-Hartman
Quoted 46 lines · Andreas Färber
Quoted 45 lines

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 12 lines · Andreas Färber
Quoted 11 lines

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Quoted 6 lines

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Seconded. For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs. People who know can check /sys/bus/soc/devices/soc0, but ordinary users will at most upload dmesg output to a Bugzilla ticket.

And it was precisely info-level boot output from the Amlogic GX driver that made me aware of this common framework and inspired me to later contribute such a driver elsewhere myself. That's not a bad effect. ;)

So if anything, we should standardize that output and move it into soc_device_register(): "<family> <soc_id> [rev <revision>] detected" with suitable NULL checks? (what I did above for Realtek, minus hex)

The info level seems correct to me - it allows people to use a different log_level if they only care about warnings or errors on the console; it's not debug info for that driver, except in my case the accompanying hex numbers that I'd be happy to drop in favor of standardization.

Quoted 12 lines

Another aspect here is that the overall amount of soc_device_register() users is just an estimated one third above the analysis shared before. In particular server platforms, be it arm64 or x86_64, that potentially have more than one SoC integrated in a multi-socket configuration, don't feed into this soc bus at all! Therefore my comment that dev_info()-printed "soc soc0:" is kind of useless if there's only one SoC on those boards. I'm not aware of any tool or a more common file aggregating this information, certainly not /proc/cpuinfo and I'm not aware of any special "lssoc" tool either. And if there's no ACPI/DMI driver feeding support-relevant information into this framework to be generally useful, I don't expect the big distros to spend time on improving its usability.

lshw? That works with info from DT, sysfs, and DMI. It did have some endian bugs (written for sparc/power) last time I looked at it 5+ years ago.

Quoted 21 lines

So if we move info output into base/soc.c, we could continue using dev_info() with "soc"-grep'able prefix in the hopes that someday we'll have more than one soc device on the bus and will need to distinguish; otherwise yes, pr_info() would change the output format for Amlogic (and so would a harmonization of the text), but does anyone really care in practice? Tools shouldn't be relying on its output format, and humans will understand equally either way.

Finally, arch/arm seems unique in that it has the machine_desc mechanism that allows individual SoCs to force creating their soc_device early and using it as parent for further DT-created platform_devices. With arm64 we've lost that ability, and nios2 is not using it either.
A bad side effect (with SUSE hat on) is that this parent design pattern does not allow to build such drivers as modules, which means that distro configs using arm's multi-platform, e.g., CONFIG_ARCH_MULTI_V7 will get unnecessary code creep as new platforms get added over time (beyond the basic clk, pinctrl, tty and maybe timer).
Even if it were possible to call into a driver module that early, using it as parent would seem to imply that all the references by its children would not allow to unload the module, which I'd consider a flawed design for such an "optional" read-once driver. If we want the device hierarchy

Quoted 8 lines

to have a soc root then most DT based platforms will have a /soc DT node anyway (although no device_type = "soc") that we probably could have a device registered for in common code rather than each SoC platform handling that differently? That might then make soc_register_device() not the creator of the device (if pre-existent) but the supplier of data to that core device, which should then allow to unload the data provider with just the sysfs data disappearing until re-inserted (speeding up the develop-and-test cycle on say not-so-resource-constrained platforms).

I for one would like to for this to be consistent. Either we always have an SoC device parent or never. I wouldn't decide based on having a DT node or not either. Generally, we should always have MMIO devices under a bus node and perhaps more than one, but that doesn't always happen. I think building the drivers as modules is a good reason to do away with the parent device.

It would also allow getting rid of remaining of_platform_default_populate calls in arm platforms except for auxdata cases. Pretty much that's the ones you list below in arch/arm/.

Quoted 7 lines

On the other hand, one might argue that such information should just be parsed by EBBR-conformant bootloaders and be passed to the kernel via standard UEFI interfaces and DMI tables. But I'm not aware of Barebox having implemented any of that yet, and even for U-Boot (e.g., Realtek based consumer devices...) not everyone has the GPL sources or tools to update their bootloader. So, having the kernel we control gather information relevant to kernel developers does make some sense to me.

UEFI and DMI are orthogonal, right. You can't expect DMI on a UEFI+DT system.

Rob

Quoted 50 lines

Thoughts?

Regards,
Andreas

P.S. Sorry that a seemingly trivial one-line 0-day fix derailed into this fundamental use case discussion.

arch/arm/mach-ep93xx/core.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-imx/cpu.c:        soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-mvebu/mvebu-soc-id.c:     soc_dev =
soc_device_register(soc_dev_attr);
arch/arm/mach-mxs/mach-mxs.c:   soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-omap2/id.c:       soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-tegra/tegra.c:    struct device *parent =
tegra_soc_device_register();
arch/arm/mach-zynq/common.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/nios2/platform/platform.c:         soc_dev =

soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-gx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-mx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);

drivers/soc/atmel/soc.c:        soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/bcm/brcmstb/common.c:       soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/fsl/guts.c: soc_dev = soc_device_register(&soc_dev_attr);
drivers/soc/imx/soc-imx-scu.c:  soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/imx/soc-imx8.c:     soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/qcom/socinfo.c:     qs->soc_dev =

soc_device_register(&qs->attr);
drivers/soc/realtek/chip.c: soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/renesas/renesas-soc.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/samsung/exynos-chipid.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/tegra/fuse/fuse-tegra.c: dev = soc_device_register(attr);
drivers/soc/ux500/ux500-soc-id.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-integrator.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-realview.c: soc_dev =
soc_device_register(soc_dev_attr);

--
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)

#15 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Am 14.11.19 um 23:09 schrieb Rob Herring:

Quoted 10 lines · Andreas Färber
Quoted 7 lines

On the other hand, one might argue that such information should just be parsed by EBBR-conformant bootloaders and be passed to the kernel via standard UEFI interfaces and DMI tables. But I'm not aware of Barebox having implemented any of that yet, and even for U-Boot (e.g., Realtek based consumer devices...) not everyone has the GPL sources or tools to update their bootloader. So, having the kernel we control gather information relevant to kernel developers does make some sense to me.

UEFI and DMI are orthogonal, right. You can't expect DMI on a UEFI+DT system.

Not sure, that's why I CC'ed the EBBR folks for input. If it's not mandatory today, the next revision of EBBR could just require it - if that's the unified way between SBBR and EBBR. Little point in doing it only for EBBR.

On my UEFI+DT based Raspberry Pi 3 Model B I do see it, note the non-filled Processor Information delivered by U-Boot:

raspi3:~ # dmidecode

# dmidecode 3.2
Getting SMBIOS data from sysfs.
SMBIOS 3.0 present.
7 structures occupying 253 bytes.
Table at 0x3CB3E020.

Handle 0x0000, DMI type 0, 24 bytes
BIOS Information
Vendor: U-Boot
Version: 2019.10
Release Date: 10/26/2019
ROM Size: 64 kB
Characteristics:
PCI is supported
BIOS is upgradeable
Selectable boot is supported
I2O boot is supported
Targeted content distribution is supported

Handle 0x0001, DMI type 1, 27 bytes
System Information
Manufacturer: raspberrypi
Product Name: rpi
Version: Not Specified
Serial Number: 00000000********
UUID: 30303030-3030-3030-6437-623461336666 Wake-up Type: Reserved
SKU Number: Not Specified
Family: Not Specified

Handle 0x0002, DMI type 2, 14 bytes
Base Board Information
Manufacturer: raspberrypi
Product Name: rpi
Version: Not Specified
Serial Number: Not Specified
Asset Tag: Not Specified
Features:
Board is a hosting board
Location In Chassis: Not Specified
Chassis Handle: 0x0000
Type: Motherboard

Handle 0x0003, DMI type 3, 21 bytes
Chassis Information
Manufacturer: raspberrypi
Type: Desktop
Lock: Not Present
Version: Not Specified
Serial Number: Not Specified
Asset Tag: Not Specified
Boot-up State: Safe
Power Supply State: Safe
Thermal State: Safe
Security Status: None
OEM Information: 0x00000000
Height: Unspecified
Number Of Power Cords: Unspecified
Contained Elements: 0

Handle 0x0004, DMI type 4, 48 bytes
Processor Information
Socket Designation: Not Specified
Type: Central Processor
Family: Unknown
Manufacturer: Unknown
ID: 00 00 00 00 00 00 00 00
Version: Unknown
Voltage: Unknown
External Clock: Unknown
Max Speed: Unknown
Current Speed: Unknown
Status: Unpopulated
Upgrade: None
L1 Cache Handle: Not Provided
L2 Cache Handle: Not Provided
L3 Cache Handle: Not Provided
Serial Number: Not Specified
Asset Tag: Not Specified
Part Number: Not Specified
Characteristics: None

Handle 0x0005, DMI type 32, 11 bytes
System Boot Information
Status: No errors detected

Handle 0x0006, DMI type 127, 4 bytes
End Of Table

Regards,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#16 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

In reply to Rob Herring

Am 14.11.19 um 23:09 schrieb Rob Herring:

Quoted 23 lines · Andreas Färber
Quoted 13 lines

Finally, arch/arm seems unique in that it has the machine_desc mechanism that allows individual SoCs to force creating their soc_device early and using it as parent for further DT-created platform_devices. With arm64 we've lost that ability, and nios2 is not using it either.
A bad side effect (with SUSE hat on) is that this parent design pattern does not allow to build such drivers as modules, which means that distro configs using arm's multi-platform, e.g., CONFIG_ARCH_MULTI_V7 will get unnecessary code creep as new platforms get added over time (beyond the basic clk, pinctrl, tty and maybe timer).
Even if it were possible to call into a driver module that early, using it as parent would seem to imply that all the references by its children would not allow to unload the module, which I'd consider a flawed design for such an "optional" read-once driver. If we want the device hierarchy

Quoted 8 lines

to have a soc root then most DT based platforms will have a /soc DT node anyway (although no device_type = "soc") that we probably could have a device registered for in common code rather than each SoC platform handling that differently? That might then make soc_register_device() not the creator of the device (if pre-existent) but the supplier of data to that core device, which should then allow to unload the data provider with just the sysfs data disappearing until re-inserted (speeding up the develop-and-test cycle on say not-so-resource-constrained platforms).

Quoted 3 lines

I for one would like to for this to be consistent. Either we always have an SoC device parent or never. I wouldn't decide based on having a DT node or not either.

Sure, if we can always be consistent, that might be best.

Where I was coming from here is that, if we're not supposed to use soc_device_to_device(), then we have no way to associate an of_node with the soc_device from the outside (and nobody was doing it today, as per my analysis). We'd either need a new helper of_soc_device_register() with additional argument for the node, or it would need to be done entirely in the infrastructure as I suggested, be it by looking for one hardcoded /soc node or for nodes with device_type = "soc".

Rob, in light of this discussion, should we start adding the latter property for new DTs such as my new Realtek SoCs, or was there a reason this has not been used consistently outside of powerpc and nios2? Intel/Altera appear to be the only two in arm64, with only three more in arm, none in mips.

(BTW my assumption was that we don't follow the booting-without-of.txt documented schema of soc<SOCname> so that we can share .dtsi across differently named SoCs, is that correct?)

Quoted 8 lines

Generally, we should always have MMIO devices
under a bus node and perhaps more than one, but that doesn't always happen. I think building the drivers as modules is a good reason to do away with the parent device.

It would also allow getting rid of remaining of_platform_default_populate calls in arm platforms except for auxdata cases. Pretty much that's the ones you list below in arch/arm/.

The majority was indeed passing in NULL, so yeah, we might clean that up, if someone could come up with a plan of where/how to implement it.

Cheers,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#17 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

In reply to Andreas Färber
Quoted 64 lines · Andreas Färber

Am 12.11.19 um 08:29 schrieb Uwe Kleine-König:

Quoted 56 lines · Greg Kroah-Hartman
Quoted 46 lines · Andreas Färber
Quoted 45 lines

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 12 lines · Andreas Färber
Quoted 11 lines

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Quoted 6 lines

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Quoted 7 lines

Seconded. For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs.

Hopefully we never had the case where needed to use the soc info in drivers, but now we have one and having such infrastructure already in-place will help.

Renesas platforms makes a extensive usage of the soc info infrastructure to figure out plenty of HW parameters at runtime and lower their DT changes.

Neil

Quoted 111 lines

People who know can check /sys/bus/soc/devices/soc0, but ordinary users will at most upload dmesg output to a Bugzilla ticket.

And it was precisely info-level boot output from the Amlogic GX driver that made me aware of this common framework and inspired me to later contribute such a driver elsewhere myself. That's not a bad effect. ;)

So if anything, we should standardize that output and move it into soc_device_register(): "<family> <soc_id> [rev <revision>] detected" with suitable NULL checks? (what I did above for Realtek, minus hex)

The info level seems correct to me - it allows people to use a different log_level if they only care about warnings or errors on the console; it's not debug info for that driver, except in my case the accompanying hex numbers that I'd be happy to drop in favor of standardization.

Another aspect here is that the overall amount of soc_device_register() users is just an estimated one third above the analysis shared before. In particular server platforms, be it arm64 or x86_64, that potentially have more than one SoC integrated in a multi-socket configuration, don't feed into this soc bus at all! Therefore my comment that dev_info()-printed "soc soc0:" is kind of useless if there's only one SoC on those boards. I'm not aware of any tool or a more common file aggregating this information, certainly not /proc/cpuinfo and I'm not aware of any special "lssoc" tool either. And if there's no ACPI/DMI driver feeding support-relevant information into this framework to be generally useful, I don't expect the big distros to spend time on improving its usability.

So if we move info output into base/soc.c, we could continue using dev_info() with "soc"-grep'able prefix in the hopes that someday we'll have more than one soc device on the bus and will need to distinguish; otherwise yes, pr_info() would change the output format for Amlogic (and so would a harmonization of the text), but does anyone really care in practice? Tools shouldn't be relying on its output format, and humans will understand equally either way.

Finally, arch/arm seems unique in that it has the machine_desc mechanism that allows individual SoCs to force creating their soc_device early and using it as parent for further DT-created platform_devices. With arm64 we've lost that ability, and nios2 is not using it either.
A bad side effect (with SUSE hat on) is that this parent design pattern does not allow to build such drivers as modules, which means that distro configs using arm's multi-platform, e.g., CONFIG_ARCH_MULTI_V7 will get unnecessary code creep as new platforms get added over time (beyond the basic clk, pinctrl, tty and maybe timer).
Even if it were possible to call into a driver module that early, using it as parent would seem to imply that all the references by its children would not allow to unload the module, which I'd consider a flawed design for such an "optional" read-once driver. If we want the device hierarchy to have a soc root then most DT based platforms will have a /soc DT node anyway (although no device_type = "soc") that we probably could have a device registered for in common code rather than each SoC platform handling that differently? That might then make soc_register_device() not the creator of the device (if pre-existent) but the supplier of data to that core device, which should then allow to unload the data provider with just the sysfs data disappearing until re-inserted (speeding up the develop-and-test cycle on say not-so-resource-constrained platforms).

On the other hand, one might argue that such information should just be parsed by EBBR-conformant bootloaders and be passed to the kernel via standard UEFI interfaces and DMI tables. But I'm not aware of Barebox having implemented any of that yet, and even for U-Boot (e.g., Realtek based consumer devices...) not everyone has the GPL sources or tools to update their bootloader. So, having the kernel we control gather information relevant to kernel developers does make some sense to me.

Thoughts?

Regards,
Andreas

P.S. Sorry that a seemingly trivial one-line 0-day fix derailed into this fundamental use case discussion.

arch/arm/mach-ep93xx/core.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-imx/cpu.c:        soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-mvebu/mvebu-soc-id.c:     soc_dev =
soc_device_register(soc_dev_attr);
arch/arm/mach-mxs/mach-mxs.c:   soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-omap2/id.c:       soc_dev = soc_device_register(soc_dev_attr);
arch/arm/mach-tegra/tegra.c:    struct device *parent =
tegra_soc_device_register();
arch/arm/mach-zynq/common.c:    soc_dev = soc_device_register(soc_dev_attr);
arch/nios2/platform/platform.c:         soc_dev =

soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-gx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/amlogic/meson-mx-socinfo.c: soc_dev =
soc_device_register(soc_dev_attr);

drivers/soc/atmel/soc.c:        soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/bcm/brcmstb/common.c:       soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/fsl/guts.c: soc_dev = soc_device_register(&soc_dev_attr);
drivers/soc/imx/soc-imx-scu.c:  soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/imx/soc-imx8.c:     soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/qcom/socinfo.c:     qs->soc_dev =

soc_device_register(&qs->attr);
drivers/soc/realtek/chip.c: soc_dev = soc_device_register(soc_dev_attr);
drivers/soc/renesas/renesas-soc.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/samsung/exynos-chipid.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/tegra/fuse/fuse-tegra.c: dev = soc_device_register(attr);
drivers/soc/ux500/ux500-soc-id.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-integrator.c: soc_dev =
soc_device_register(soc_dev_attr);
drivers/soc/versatile/soc-realview.c: soc_dev =
soc_device_register(soc_dev_attr);

#18 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Hi Neil,

Quoted 73 lines · Neil Armstrong
Quoted 64 lines · Andreas Färber

Am 12.11.19 um 08:29 schrieb Uwe Kleine-König:

Quoted 56 lines · Greg Kroah-Hartman
Quoted 46 lines · Andreas Färber
Quoted 45 lines

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 12 lines · Andreas Färber
Quoted 11 lines

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Quoted 6 lines

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Quoted 7 lines

Seconded. For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs.

Quoted 5 lines

Hopefully we never had the case where needed to use the soc info in drivers, but now we have one and having such infrastructure already in-place will help.

Renesas platforms makes a extensive usage of the soc info infrastructure to figure out plenty of HW parameters at runtime and lower their DT changes.

We do our best to use it solely for detecting quirks in early SoC revisions.

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#19 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Hi Geert,

Am 15.11.19 um 09:58 schrieb Geert Uytterhoeven:

Quoted 10 lines · Neil Armstrong
Quoted 9 lines · Andreas Färber
Quoted 7 lines

For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs.

Quoted 7 lines
Quoted 5 lines

Hopefully we never had the case where needed to use the soc info in drivers, but now we have one and having such infrastructure already in-place will help.

Renesas platforms makes a extensive usage of the soc info infrastructure to figure out plenty of HW parameters at runtime and lower their DT changes.

We do our best to use it solely for detecting quirks in early SoC revisions.

Got a pointer? I fail to immediately understand how sysfs would help drivers (as opposed to userspace) detect quirks: Parsing strings back doesn't sound efficient, and I don't see you exporting any custom APIs in drivers/soc/renesas/renesas-soc.c?

Regards,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#20 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Hi Andreas,

Quoted 17 lines · Andreas Färber

Am 15.11.19 um 09:58 schrieb Geert Uytterhoeven:

Quoted 16 lines · Neil Armstrong
Quoted 14 lines · Andreas Färber
Quoted 7 lines

For example, RTD1295 will support LSADC only from revision B00 on (and it's not the first time I'm seeing such things in the industry). So if a user complains, it will be helpful to see that information.

Referencing your Amlogic review, with all due respect for its authors, the common framework here just lets that information evaporate into the deeps of sysfs.

Hopefully we never had the case where needed to use the soc info in drivers, but now we have one and having such infrastructure already in-place will help.

Renesas platforms makes a extensive usage of the soc info infrastructure to figure out plenty of HW parameters at runtime and lower their DT changes.

Quoted 6 lines
Quoted 1 line

We do our best to use it solely for detecting quirks in early SoC revisions.

Got a pointer? I fail to immediately understand how sysfs would help drivers (as opposed to userspace) detect quirks: Parsing strings back doesn't sound efficient, and I don't see you exporting any custom APIs in drivers/soc/renesas/renesas-soc.c?

We use soc_device_match(), inside kernel drivers.
Exposure through sysfs is a side-effect of using soc_device_register(), and welcomed, as it allows the user to find out quickly which SoC and revision is being used.

FTR, lshw (Ubuntu 18.04 has v2.18, which does seem to be the latest upstream version) does not parse /sys/devices/soc0/.

Gr{oetje,eeting}s,

Geert

Signature · 5 lines
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- [email protected]

In personal conversations with technical people, I call myself a hacker. But
when I'm talking to journalists I just say "programmer" or something like that.
                                -- Linus Torvalds
#21 lore raw

Re: Sense of soc bus? (was: [PATCH] base: soc: Export soc_device_to_device() helper)

Quoted 9 lines · Geert Uytterhoeven
Quoted 7 lines · Andreas Färber

Am 15.11.19 um 09:58 schrieb Geert Uytterhoeven:

Quoted 1 line

We do our best to use it solely for detecting quirks in early SoC revisions.

Got a pointer? I fail to immediately understand how sysfs would help drivers (as opposed to userspace) detect quirks: Parsing strings back doesn't sound efficient, and I don't see you exporting any custom APIs in drivers/soc/renesas/renesas-soc.c?

Quoted 4 lines

We use soc_device_match(), inside kernel drivers.
Exposure through sysfs is a side-effect of using soc_device_register(), and welcomed, as it allows the user to find out quickly which SoC and revision is being used.

For the omap variants too, we've so far gotten away with early SoC detection for platform code, and then use soc_device_match() in few cases for drivers at probe time if needed.

Regards,

Tony

#22 lore raw

Re: [PATCH] base: soc: Export soc_device_to_device() helper

In reply to Uwe Kleine-König
Quoted 58 lines · Uwe Kleine-König
Quoted 48 lines · Greg Kroah-Hartman
Quoted 47 lines · Andreas Färber

Am 11.11.19 um 07:40 schrieb Greg Kroah-Hartman:

Quoted 14 lines · Andreas Färber
Quoted 13 lines

Hi Greg,

Am 11.11.19 um 06:27 schrieb Greg Kroah-Hartman:

Quoted 9 lines · Andreas Färber
Quoted 3 lines

Use of soc_device_to_device() in driver modules causes a build failure. Given that the helper is nicely documented in include/linux/sys_soc.h, let's export it as GPL symbol.

I thought we were fixing the soc drivers to not need this. What happened to that effort? I thought I had patches in my tree (or someone's tree) that did some of this work already, such that this symbol isn't needed anymore.

Quoted 8 lines
Quoted 3 lines

I do still see this function used in next-20191108 in drivers/soc/.

I'll be happy to adjust my RFC driver if someone points me to how!

Look at c31e73121f4c ("base: soc: Handle custom soc information sysfs entries") for how you can just use the default attributes for the soc to create the needed sysfs files, instead of having to do it "by hand" which is racy and incorrect.

Unrelated.

Quoted 7 lines
Quoted 4 lines

Given the current struct layout, a type cast might work (but ugly). Or if we stay with my current RFC driver design, we could use the platform_device instead of the soc_device (which would clutter the screen more than "soc soc0:") or resort to pr_info() w/o device.

Ick, no, don't cast blindly. What do you need the pointer for? Is this for in-tree code?

No, an RFC patchset: https://patchwork.kernel.org/cover/11224261/

As I indicated above, I used it for a dev_info(), which I can easily avoid by using pr_info() instead:

drivers/soc/realtek/chip.c +0 -0
index e5078c6731fd..f9380e831659 100644
@@ -178,8 +178,7 @@ static int rtd_soc_probe(struct platform_device *pdev)
 
        platform_set_drvdata(pdev, soc_dev);
 
Quoted 8 lines

drivers/soc/realtek/chip.c static int rtd_soc_probe(struct platform_device *pdev) · line 183

Quoted 5 lines
drivers/soc/realtek/chip.c +1 -2
-       dev_info(soc_device_to_device(soc_dev),
-               "%s %s (0x%08x) rev %s (0x%08x) detected\n",
+       pr_info("%s %s (0x%08x) rev %s (0x%08x) detected\n",
                soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
                soc_dev_attr->revision, chip_rev);

First off, the driver should not be spitting out noise for when all goes well like this :)

Quoted 6 lines

I didn't follow the discussion closely, but I think I want to object here a bit. While I agree that each driver emitting some stuff to the log buffer is hardly helpful, seeing the exact SoC details is indeed useful at times. With my Debian kernel team member hat on, I'd say keep this information. This way the SoC details make it into kernel bug reports without effort on our side.

Right. From my angle we are starting to be way too aggressive with the point about not printing information to the kernel log. In only a small set of cases does this actually cause an issue i.e. with platforms containing so many devices that printing information from each of them does significantly increase boot times. In my world of small electronics I've been greatly hindered by the lack of information, such that it has cost days of engineering trying to track down fictitious bugs and the like.

For platforms where printing useful information culminates in negative effects, perhaps simply lower their log level, rather than suffocate all platforms.

Signature · 4 lines
Lee Jones [李琼斯]
Linaro Services Technical Lead
Linaro.org │ Open source software for ARM SoCs
Follow Linaro: Facebook | Twitter | Blog
maintainer
#23 lore raw

RE: [RFC 02/11] soc: Add Realtek chip info driver for RTD1195 and RTD1295

In reply to Andreas Färber

Add Stanley Chang for review.

Quoted 236 lines

Add a soc bus driver to print chip model and revision details.

Revisions from downstream drivers/soc/realtek/rtd{119x,129x}/rtk_chip.c.

---
Naming: What to call the family vs. soc_id?

 drivers/soc/Kconfig          |   1 +
 drivers/soc/Makefile         |   1 +
 drivers/soc/realtek/Kconfig  |  13 ++++
 drivers/soc/realtek/Makefile |   2 +
 drivers/soc/realtek/chip.c   | 164
+++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 181 insertions(+)
 create mode 100644 drivers/soc/realtek/Kconfig  create mode 100644

drivers/soc/realtek/Makefile create mode 100644 drivers/soc/realtek/chip.c

drivers/soc/Kconfig index renamed from drivers/soc/Kconfig +0 -0

833e04a7835c..06ae9d97321c 100644

drivers/soc/Kconfig +1 -0
@@ -11,6 +11,7 @@ source "drivers/soc/imx/Kconfig"
 source "drivers/soc/ixp4xx/Kconfig"
 source "drivers/soc/mediatek/Kconfig"
 source "drivers/soc/qcom/Kconfig"
+source "drivers/soc/realtek/Kconfig"
 source "drivers/soc/renesas/Kconfig"
 source "drivers/soc/rockchip/Kconfig"
 source "drivers/soc/samsung/Kconfig"
drivers/soc/Makefile index renamed from drivers/soc/Makefile +0 -0

2ec355003524..1d55d838a342 100644

drivers/soc/Makefile +1 -0
@@ -17,6 +17,7 @@ obj-$(CONFIG_SOC_XWAY)		+= lantiq/
 obj-y				+= mediatek/
 obj-y				+= amlogic/
 obj-y				+= qcom/
+obj-y				+= realtek/
 obj-y				+= renesas/
 obj-$(CONFIG_ARCH_ROCKCHIP)	+= rockchip/
 obj-$(CONFIG_SOC_SAMSUNG)	+= samsung/
drivers/soc/realtek/Kconfig new file renamed from drivers/soc/realtek/Kconfig +0 -0

mode 100644 index 000000000000..be75c1889c61

drivers/soc/realtek/Kconfig new file +13 -0
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0-or-later if ARCH_REALTEK ||
+COMPILE_TEST
+
+config REALTEK_SOC
+	tristate "Realtek chip info"
+	default ARCH_REALTEK
+	select SOC_BUS
+	help
+	  Say 'y' here to enable support for SoC info on Realtek RTD1195 and
+	  RTD1295 SoC families.
+	  If unsure, say 'n'.
+
+endif
drivers/soc/realtek/Makefile new renamed from drivers/soc/realtek/Makefile +0 -0

file mode 100644 index 000000000000..49900273905b

drivers/soc/realtek/Makefile new file +2 -0
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+obj-$(CONFIG_REALTEK_SOC) += chip.o
drivers/soc/realtek/chip.c new file renamed from drivers/soc/realtek/chip.c +0 -0

mode 100644 index 000000000000..9d13422e9936

drivers/soc/realtek/chip.c new file +160 -1
@@ -0,0 +1,164 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Realtek System-on-Chip info
+ *
+ * Copyright (c) 2017-2019 Andreas Färber  */
+
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/sys_soc.h>
+
+#define REG_CHIP_ID	0x0
+#define REG_CHIP_REV	0x4
+
+struct rtd_soc_revision {
+	const char *name;
+	u32 chip_rev;
+};
+
+static const struct rtd_soc_revision rtd1195_revisions[] = {
+	{ "A", 0x00000000 },
+	{ "B", 0x00010000 },
+	{ "C", 0x00020000 },
+	{ "D", 0x00030000 },
+	{ }
+};
+
+static const struct rtd_soc_revision rtd1295_revisions[] = {
+	{ "A00", 0x00000000 },
+	{ "A01", 0x00010000 },
+	{ "B00", 0x00020000 },
+	{ "B01", 0x00030000 },
+	{ }
+};
+
+struct rtd_soc {
+	u32 chip_id;
+	const char *family;
+	const char *(*get_name)(struct device *dev, const struct rtd_soc *s);
+	const struct rtd_soc_revision *revisions;
+	const char *codename;
+};
+
+static const char *default_name(struct device *dev, const struct
+rtd_soc *s) {
+	return s->family;
+}
+
+static const struct rtd_soc rtd_soc_families[] = {
+	{ 0x00006329, "RTD1195", default_name, rtd1195_revisions, "Phoenix" },
+	{ 0x00006421, "RTD1295", default_name, rtd1295_revisions, "Kylin" },
+};
+
+static const struct rtd_soc *rtd_soc_by_chip_id(u32 chip_id) {
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(rtd_soc_families); i++) {
+		const struct rtd_soc *family = &rtd_soc_families[i];
+
+		if (family->chip_id == chip_id)
+			return family;
+	}
+	return NULL;
+}
+
+static const char *rtd_soc_rev(const struct rtd_soc *family, u32
+chip_rev) {
+	if (family) {
+		const struct rtd_soc_revision *rev = family->revisions;
+
+		while (rev && rev->name) {
+			if (rev->chip_rev == chip_rev)
+				return rev->name;
+			rev++;
+		}
+	}
+	return "unknown";
+}
+
+static int rtd_soc_probe(struct platform_device *pdev) {
+	const struct rtd_soc *s;
+	struct soc_device_attribute *soc_dev_attr;
+	struct soc_device *soc_dev;
+	struct device_node *node;
+	void __iomem *base;
+	u32 chip_id, chip_rev;
+
+	base = of_iomap(pdev->dev.of_node, 0);
+	if (!base)
+		return -ENODEV;
+
+	soc_dev_attr = kzalloc(sizeof(*soc_dev_attr), GFP_KERNEL);
+	if (!soc_dev_attr)
+		return -ENOMEM;
+
+	chip_id  = readl_relaxed(base + REG_CHIP_ID);
+	chip_rev = readl_relaxed(base + REG_CHIP_REV);
+
+	node = of_find_node_by_path("/");
+	of_property_read_string(node, "model", &soc_dev_attr->machine);
+	of_node_put(node);
+
+	s = rtd_soc_by_chip_id(chip_id);
+
+	soc_dev_attr->family = kasprintf(GFP_KERNEL, "Realtek %s",
+		(s && s->codename) ? s->codename :
+		((s && s->family) ? s->family : "Digital Home Center"));
+
+	if (likely(s && s->get_name))
+		soc_dev_attr->soc_id = s->get_name(&pdev->dev, s);
+	else
+		soc_dev_attr->soc_id = "unknown";
+
+	soc_dev_attr->revision = rtd_soc_rev(s, chip_rev);
+
+	soc_dev = soc_device_register(soc_dev_attr);
+	if (IS_ERR(soc_dev)) {
+		kfree(soc_dev_attr->family);
+		kfree(soc_dev_attr);
+		return PTR_ERR(soc_dev);
+	}
+
+	platform_set_drvdata(pdev, soc_dev);
+
+	dev_info(soc_device_to_device(soc_dev),
+		"%s %s (0x%08x) rev %s (0x%08x) detected\n",
+		soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
+		soc_dev_attr->revision, chip_rev);
+
+	return 0;
+}
+
+static int rtd_soc_remove(struct platform_device *pdev) {
+	struct soc_device *soc_dev = platform_get_drvdata(pdev);
+
+	soc_device_unregister(soc_dev);
+
+	return 0;
+}
+
+static const struct of_device_id rtd_soc_dt_ids[] = {
+	 { .compatible = "realtek,rtd1195-chip" },
+	 { }
+};
+
+static struct platform_driver rtd_soc_driver = {
+	.probe = rtd_soc_probe,
+	.remove = rtd_soc_remove,
+	.driver = {
+		.name = "rtd1195-soc",
+		.of_match_table	= rtd_soc_dt_ids,
+	},
+};
+module_platform_driver(rtd_soc_driver);
+
+MODULE_DESCRIPTION("Realtek SoC identification");
+MODULE_LICENSE("GPL");
--

2.16.4

Quoted 6 lines

_______________________________________________
linux-realtek-soc mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-realtek-soc

------Please consider the environment before printing this e-mail.

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#24 lore raw

Re: [RFC 02/11] soc: Add Realtek chip info driver for RTD1195 and RTD1295

Am 02.01.20 um 15:29 schrieb James Tai:

Quoted 1 line

Add Stanley Chang for review.

Did you forget to CC him?

Note that this series needs updates once we apply my syscon patches.

Quoted 102 lines
Quoted 102 lines

Add a soc bus driver to print chip model and revision details.

Revisions from downstream drivers/soc/realtek/rtd{119x,129x}/rtk_chip.c.

---
Naming: What to call the family vs. soc_id?

  drivers/soc/Kconfig          |   1 +
  drivers/soc/Makefile         |   1 +
  drivers/soc/realtek/Kconfig  |  13 ++++
  drivers/soc/realtek/Makefile |   2 +
  drivers/soc/realtek/chip.c   | 164
+++++++++++++++++++++++++++++++++++++++++++
  5 files changed, 181 insertions(+)

create mode 100644 drivers/soc/realtek/Kconfig create mode 100644
drivers/soc/realtek/Makefile create mode 100644 drivers/soc/realtek/chip.c

drivers/soc/Kconfig index renamed from drivers/soc/Kconfig +0 -0

833e04a7835c..06ae9d97321c 100644

drivers/soc/Kconfig +1 -0
@@ -11,6 +11,7 @@ source "drivers/soc/imx/Kconfig"
  source "drivers/soc/ixp4xx/Kconfig"
  source "drivers/soc/mediatek/Kconfig"
  source "drivers/soc/qcom/Kconfig"
+source "drivers/soc/realtek/Kconfig"
  source "drivers/soc/renesas/Kconfig"
  source "drivers/soc/rockchip/Kconfig"
  source "drivers/soc/samsung/Kconfig"
drivers/soc/Makefile index renamed from drivers/soc/Makefile +0 -0

2ec355003524..1d55d838a342 100644

drivers/soc/Makefile +1 -0
@@ -17,6 +17,7 @@ obj-$(CONFIG_SOC_XWAY)		+= lantiq/
  obj-y				+= mediatek/
  obj-y				+= amlogic/
  obj-y				+= qcom/
+obj-y				+= realtek/
  obj-y				+= renesas/
  obj-$(CONFIG_ARCH_ROCKCHIP)	+= rockchip/
  obj-$(CONFIG_SOC_SAMSUNG)	+= samsung/
drivers/soc/realtek/Kconfig new file renamed from drivers/soc/realtek/Kconfig +0 -0

mode 100644 index 000000000000..be75c1889c61

drivers/soc/realtek/Kconfig new file +13 -0
@@ -0,0 +1,13 @@
+# SPDX-License-Identifier: GPL-2.0-or-later if ARCH_REALTEK ||
+COMPILE_TEST
+
+config REALTEK_SOC
+	tristate "Realtek chip info"
+	default ARCH_REALTEK
+	select SOC_BUS
+	help
+	  Say 'y' here to enable support for SoC info on Realtek RTD1195 and
+	  RTD1295 SoC families.
+	  If unsure, say 'n'.
+
+endif
drivers/soc/realtek/Makefile new renamed from drivers/soc/realtek/Makefile +0 -0

file mode 100644 index 000000000000..49900273905b

drivers/soc/realtek/Makefile new file +2 -0
@@ -0,0 +1,2 @@
+# SPDX-License-Identifier: GPL-2.0-or-later
+obj-$(CONFIG_REALTEK_SOC) += chip.o
drivers/soc/realtek/chip.c new file renamed from drivers/soc/realtek/chip.c +0 -0

mode 100644 index 000000000000..9d13422e9936

drivers/soc/realtek/chip.c new file +29 -0
@@ -0,0 +1,164 @@
+// SPDX-License-Identifier: GPL-2.0-or-later
+/*
+ * Realtek System-on-Chip info
+ *
+ * Copyright (c) 2017-2019 Andreas Färber  */
+
+#include <linux/io.h>
+#include <linux/module.h>
+#include <linux/of.h>
+#include <linux/of_address.h>
+#include <linux/platform_device.h>
+#include <linux/slab.h>
+#include <linux/sys_soc.h>
+
+#define REG_CHIP_ID	0x0
+#define REG_CHIP_REV	0x4
+
+struct rtd_soc_revision {
+	const char *name;
+	u32 chip_rev;
+};
+
+static const struct rtd_soc_revision rtd1195_revisions[] = {
+	{ "A", 0x00000000 },
+	{ "B", 0x00010000 },
+	{ "C", 0x00020000 },
+	{ "D", 0x00030000 },
+	{ }
+};
Quoted 8 lines

drivers/soc/realtek/chip.c line 38

Quoted 8 lines
drivers/soc/realtek/chip.c new file +8 -0
+
+static const struct rtd_soc_revision rtd1295_revisions[] = {
+	{ "A00", 0x00000000 },
+	{ "A01", 0x00010000 },
+	{ "B00", 0x00020000 },
+	{ "B01", 0x00030000 },
+	{ }
+};

I believe the lower 16 bits are reserved, so we should only be comparing the upper 16.

Quoted 10 lines
Quoted 10 lines
drivers/soc/realtek/chip.c new file +10 -0
+
+struct rtd_soc {
+	u32 chip_id;
+	const char *family;
+	const char *(*get_name)(struct device *dev, const struct rtd_soc *s);
+	const struct rtd_soc_revision *revisions;
+	const char *codename;
+};
+
+static const char *default_name(struct device *dev, const struct
Quoted 8 lines

drivers/soc/realtek/chip.c line 56

Quoted 8 lines
drivers/soc/realtek/chip.c new file +8 -0
+rtd_soc *s) {
+	return s->family;
+}
+
+static const struct rtd_soc rtd_soc_families[] = {
+	{ 0x00006329, "RTD1195", default_name, rtd1195_revisions, "Phoenix" },
+	{ 0x00006421, "RTD1295", default_name, rtd1295_revisions, "Kylin" },
+};

Similarly here I believe the upper 16 bits are reserved, so we should only be comparing the lower 16, which should make it easier to stay within 80 characters per line.

Regards,
Andreas

Quoted 115 lines
Quoted 115 lines
drivers/soc/realtek/chip.c new file +105 -1
+
+static const struct rtd_soc *rtd_soc_by_chip_id(u32 chip_id) {
+	int i;
+
+	for (i = 0; i < ARRAY_SIZE(rtd_soc_families); i++) {
+		const struct rtd_soc *family = &rtd_soc_families[i];
+
+		if (family->chip_id == chip_id)
+			return family;
+	}
+	return NULL;
+}
+
+static const char *rtd_soc_rev(const struct rtd_soc *family, u32
+chip_rev) {
+	if (family) {
+		const struct rtd_soc_revision *rev = family->revisions;
+
+		while (rev && rev->name) {
+			if (rev->chip_rev == chip_rev)
+				return rev->name;
+			rev++;
+		}
+	}
+	return "unknown";
+}
+
+static int rtd_soc_probe(struct platform_device *pdev) {
+	const struct rtd_soc *s;
+	struct soc_device_attribute *soc_dev_attr;
+	struct soc_device *soc_dev;
+	struct device_node *node;
+	void __iomem *base;
+	u32 chip_id, chip_rev;
+
+	base = of_iomap(pdev->dev.of_node, 0);
+	if (!base)
+		return -ENODEV;
+
+	soc_dev_attr = kzalloc(sizeof(*soc_dev_attr), GFP_KERNEL);
+	if (!soc_dev_attr)
+		return -ENOMEM;
+
+	chip_id  = readl_relaxed(base + REG_CHIP_ID);
+	chip_rev = readl_relaxed(base + REG_CHIP_REV);
+
+	node = of_find_node_by_path("/");
+	of_property_read_string(node, "model", &soc_dev_attr->machine);
+	of_node_put(node);
+
+	s = rtd_soc_by_chip_id(chip_id);
+
+	soc_dev_attr->family = kasprintf(GFP_KERNEL, "Realtek %s",
+		(s && s->codename) ? s->codename :
+		((s && s->family) ? s->family : "Digital Home Center"));
+
+	if (likely(s && s->get_name))
+		soc_dev_attr->soc_id = s->get_name(&pdev->dev, s);
+	else
+		soc_dev_attr->soc_id = "unknown";
+
+	soc_dev_attr->revision = rtd_soc_rev(s, chip_rev);
+
+	soc_dev = soc_device_register(soc_dev_attr);
+	if (IS_ERR(soc_dev)) {
+		kfree(soc_dev_attr->family);
+		kfree(soc_dev_attr);
+		return PTR_ERR(soc_dev);
+	}
+
+	platform_set_drvdata(pdev, soc_dev);
+
+	dev_info(soc_device_to_device(soc_dev),
+		"%s %s (0x%08x) rev %s (0x%08x) detected\n",
+		soc_dev_attr->family, soc_dev_attr->soc_id, chip_id,
+		soc_dev_attr->revision, chip_rev);
+
+	return 0;
+}
+
+static int rtd_soc_remove(struct platform_device *pdev) {
+	struct soc_device *soc_dev = platform_get_drvdata(pdev);
+
+	soc_device_unregister(soc_dev);
+
+	return 0;
+}
+
+static const struct of_device_id rtd_soc_dt_ids[] = {
+	 { .compatible = "realtek,rtd1195-chip" },
+	 { }
+};
+
+static struct platform_driver rtd_soc_driver = {
+	.probe = rtd_soc_probe,
+	.remove = rtd_soc_remove,
+	.driver = {
+		.name = "rtd1195-soc",
+		.of_match_table	= rtd_soc_dt_ids,
+	},
+};
+module_platform_driver(rtd_soc_driver);
+
+MODULE_DESCRIPTION("Realtek SoC identification");
+MODULE_LICENSE("GPL");
--

2.16.4

_______________________________________________
linux-realtek-soc mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-realtek-soc

------Please consider the environment before printing this e-mail.

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
Tested-by Reviewed-by
#26 lore raw

RE: [RFC 02/11] soc: Add Realtek chip info driver for RTD1195 and RTD1295

In reply to James Tai

Hi Andreas,

I have tested this patch in my local platform.
And this patch is fine and it can work.

Quoted 21 lines

Add Stanley Chang for review.

Quoted 18 lines

Add a soc bus driver to print chip model and revision details.

Revisions from downstream drivers/soc/realtek/rtd{119x,129x}/rtk_chip.c.

---
Naming: What to call the family vs. soc_id?

 drivers/soc/Kconfig          |   1 +
 drivers/soc/Makefile         |   1 +
 drivers/soc/realtek/Kconfig  |  13 ++++
 drivers/soc/realtek/Makefile |   2 +
 drivers/soc/realtek/chip.c   | 164
+++++++++++++++++++++++++++++++++++++++++++
 5 files changed, 181 insertions(+)
 create mode 100644 drivers/soc/realtek/Kconfig  create mode 100644

drivers/soc/realtek/Makefile create mode 100644
drivers/soc/realtek/chip.c

Thanks,
Stanley
_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#27 lore raw

RFC 3/11 arm64: dts: realtek: rtd129x: Add chip info node

In reply to Andreas Färber

Add a DT node for chip identification.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +5 -0
index 4433114476f5..15a7c249155d 100644
@@ -84,6 +84,11 @@
 			status = "disabled";
 		};
 
+		chip-info@9801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x9801a200 0x8>;
+		};
+
 		uart1: serial@9801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x9801b200 0x100>;
Signature · 1 line
2.16.4
maintainer
#28 lore raw

RE: [RFC 03/11] arm64: dts: realtek: rtd129x: Add chip info node

Add Stanley Chang for review.

Quoted 34 lines

Add a DT node for chip identification.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +5 -1
index 4433114476f5..15a7c249155d 100644
@@ -84,6 +84,11 @@
 			status = "disabled";
 		};
 
+		chip-info@9801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x9801a200 0x8>;
+		};
+
 		uart1: serial@9801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x9801b200 0x100>;
--

2.16.4

_______________________________________________
linux-realtek-soc mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-realtek-soc

------Please consider the environment before printing this e-mail.

Tested-by Reviewed-by
#29 lore raw

RE: [RFC 03/11] arm64: dts: realtek: rtd129x: Add chip info node

Hi Andreas,

This patch is work in my platform.

Quoted 18 lines

Add Stanley Chang for review.

Quoted 16 lines

Add a DT node for chip identification.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +0 -0
index 4433114476f5..15a7c249155d 100644
@@ -84,6 +84,11 @@
 			status = "disabled";
 		};
 
Quoted 11 lines
Quoted 11 lines
arch/arm64/boot/dts/realtek/rtd129x.dtsi +5 -1
+		chip-info@9801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x9801a200 0x8>;
+		};
+
 		uart1: serial@9801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x9801b200 0x100>;
--

2.16.4

Thanks,
Stanley

maintainer Acked-by on v1 3/11
#30 lore raw

RE: [RFC 03/11] arm64: dts: realtek: rtd129x: Add chip info node

In reply to Andreas Färber
Quoted 25 lines

Add a DT node for chip identification.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +5 -1
index 4433114476f5..15a7c249155d 100644
@@ -84,6 +84,11 @@
 			status = "disabled";
 		};
 
+		chip-info@9801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x9801a200 0x8>;
+		};
+
 		uart1: serial@9801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x9801b200 0x100>;
--
Quoted 2 lines

2.16.4

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

maintainer Acked-by on v1 3/11
#31 lore raw

RE: [RFC 03/11] arm64: dts: realtek: rtd129x: Add chip info node

In reply to Andreas Färber
Quoted 25 lines

Add a DT node for chip identification.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +5 -1
index 4433114476f5..15a7c249155d 100644
@@ -84,6 +84,11 @@
 			status = "disabled";
 		};
 
+		chip-info@9801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x9801a200 0x8>;
+		};
+
 		uart1: serial@9801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x9801b200 0x100>;
--
Quoted 2 lines

2.16.4

#32 lore raw

RFC 4/11 ARM: dts: rtd1195: Add chip info node

In reply to Andreas Färber

Add a DT node for chip identification.

---
 arch/arm/boot/dts/rtd1195.dtsi | 5 +++++
 1 file changed, 5 insertions(+)
arch/arm/boot/dts/rtd1195.dtsi +5 -0
index 9e088f669bd7..dd4f3ee6be43 100644
@@ -133,6 +133,11 @@
 			status = "disabled";
 		};
 
+		chip-info@1801a200 {
+			compatible = "realtek,rtd1195-chip";
+			reg = <0x1801a200 0x8>;
+		};
+
 		uart1: serial@1801b200 {
 			compatible = "snps,dw-apb-uart";
 			reg = <0x1801b200 0x100>;
Signature · 7 lines
2.16.4


_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel
#33 lore raw

RFC 5/11 dt-bindings: soc: realtek: rtd1195-chip: Extend reg property

In reply to Andreas Färber

Allow to optionally specify a second register to identify the chip. Whether needed and which register to specify depends on the family; RTD1295 family will want the CHIP_INFO1 register.

---
A SoC specific binding would defeat the purpose of the generic Linux driver; is it possible to check the root node's compatible in an if: expression to prohibit using more than one reg on "realtek,rtd1195"?

 .../devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml  | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml +9 -1
index 565ad2419553..e431cf559b66 100644
@@ -11,13 +11,15 @@ maintainers:
 
 description: |
   The Realtek SoCs have some registers to identify the chip and revision.
+  To identify the exact model within a family, further registers are needed.
 
 properties:
   compatible:
     const: "realtek,rtd1195-chip"
 
   reg:
-    maxItems: 1
+    minItems: 1
+    maxItems: 2
 
 required:
   - compatible
@@ -29,4 +31,10 @@ examples:
         compatible = "realtek,rtd1195-chip";
         reg = <0x1801a200 0x8>;
     };
+  - |
+    chip-info@9801a200 {
+        compatible = "realtek,rtd1195-chip";
+        reg = <0x9801a200 0x8>,
+              <0x98007028 0x4>;
+    };
 ...
Signature · 1 line
2.16.4
maintainer
#34 lore raw

Re: [RFC 05/11] dt-bindings: soc: realtek: rtd1195-chip: Extend reg property

Quoted 7 lines · Andreas Färber

Allow to optionally specify a second register to identify the chip. Whether needed and which register to specify depends on the family; RTD1295 family will want the CHIP_INFO1 register.

---
A SoC specific binding would defeat the purpose of the generic Linux driver;

Why? You can map any number of compatibles to a generic driver.

Quoted 2 lines

is it possible to check the root node's compatible in an if: expression to prohibit using more than one reg on "realtek,rtd1195"?

The "rule" is different programming model, different compatible string for the block. But this looks simple enough, I don't really care.

Quoted 39 lines
 
 .../devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml  | 10 +++++++++-
 1 file changed, 9 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml +9 -1
index 565ad2419553..e431cf559b66 100644
@@ -11,13 +11,15 @@ maintainers:
 
 description: |
   The Realtek SoCs have some registers to identify the chip and revision.
+  To identify the exact model within a family, further registers are needed.
 
 properties:
   compatible:
     const: "realtek,rtd1195-chip"
 
   reg:
-    maxItems: 1
+    minItems: 1
+    maxItems: 2
 
 required:
   - compatible
@@ -29,4 +31,10 @@ examples:
         compatible = "realtek,rtd1195-chip";
         reg = <0x1801a200 0x8>;
     };
+  - |
+    chip-info@9801a200 {
+        compatible = "realtek,rtd1195-chip";
+        reg = <0x9801a200 0x8>,
+              <0x98007028 0x4>;
+    };
...
Signature · 1 line
2.16.4
#35 lore raw

Re: [RFC 05/11] dt-bindings: soc: realtek: rtd1195-chip: Extend reg property

Am 06.11.19 um 05:46 schrieb Rob Herring:

Quoted 10 lines · Andreas Färber
Quoted 7 lines

Allow to optionally specify a second register to identify the chip. Whether needed and which register to specify depends on the family; RTD1295 family will want the CHIP_INFO1 register.

---
A SoC specific binding would defeat the purpose of the generic Linux driver;

Why? You can map any number of compatibles to a generic driver.

Because the purpose of the driver is to read from the registers which chip it is. If we tell it via the compatible what it is supposed to be,
1) only the revision would need to be read, and 2) how should it react if the compatible tells it one thing and the register value another.

Also it doesn't solve the problem that we may need to extend the binding as new models emerge, or instead of just rtd1195, rtd1295, rtd1395, etc. we'd also need one for each chip, i.e., rtd1296, cf. 1) above.

Quoted 5 lines
Quoted 2 lines

is it possible to check the root node's compatible in an if: expression to prohibit using more than one reg on "realtek,rtd1195"?

The "rule" is different programming model, different compatible string for the block.

Agreed in general.

Quoted 1 line

But this looks simple enough, I don't really care.

Hope you also read the cover letter wrt syscon? That would probably obsolete this binding then and require to move the driver's logic into a module init instead for lack of dedicated compatible to bind against, like Meson does.

Regards,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)
#36 lore raw

RFC 6/11 soc: realtek: chip: Detect RTD1296

In reply to Andreas Färber

Detection logic from downstream drivers/soc/realtek/rtd129x/rtk_chip.c.

---
 drivers/soc/realtek/chip.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)
drivers/soc/realtek/chip.c +17 -1
index 9d13422e9936..ba653c097644 100644
@@ -50,9 +50,25 @@ static const char *default_name(struct device *dev, const struct rtd_soc *s)
 	return s->family;
 }
 
+static const char *rtd1295_name(struct device *dev, const struct rtd_soc *s)
+{
+	void __iomem *base;
+
+	base = of_iomap(dev->of_node, 1);
+	if (base) {
+		u32 chipinfo1 = readl_relaxed(base);
+		iounmap(base);
+		if (chipinfo1 & BIT(11)) {
+			return "RTD1296";
+		}
+	}
+
+	return "RTD1295";
+}
+
 static const struct rtd_soc rtd_soc_families[] = {
 	{ 0x00006329, "RTD1195", default_name, rtd1195_revisions, "Phoenix" },
-	{ 0x00006421, "RTD1295", default_name, rtd1295_revisions, "Kylin" },
+	{ 0x00006421, "RTD1295", rtd1295_name, rtd1295_revisions, "Kylin" },
 };
 
 static const struct rtd_soc *rtd_soc_by_chip_id(u32 chip_id)
Signature · 1 line
2.16.4
maintainer Acked-by on v1 6/11
#37 lore raw

RE: [RFC 06/11] soc: realtek: chip: Detect RTD1296

Quoted 38 lines

Detection logic from downstream drivers/soc/realtek/rtd129x/rtk_chip.c.

---
 drivers/soc/realtek/chip.c | 18 +++++++++++++++++-
 1 file changed, 17 insertions(+), 1 deletion(-)
drivers/soc/realtek/chip.c index renamed from drivers/soc/realtek/chip.c +0 -0

9d13422e9936..ba653c097644 100644

drivers/soc/realtek/chip.c +17 -1
@@ -50,9 +50,25 @@ static const char *default_name(struct device *dev,
 const struct rtd_soc *s)
 	return s->family;
 }
 
+static const char *rtd1295_name(struct device *dev, const struct
+rtd_soc *s) {
+	void __iomem *base;
+
+	base = of_iomap(dev->of_node, 1);
+	if (base) {
+		u32 chipinfo1 = readl_relaxed(base);
+		iounmap(base);
+		if (chipinfo1 & BIT(11)) {
+			return "RTD1296";
+		}
+	}
+
+	return "RTD1295";
+}
+
 static const struct rtd_soc rtd_soc_families[] = {
 	{ 0x00006329, "RTD1195", default_name, rtd1195_revisions, "Phoenix" },
-	{ 0x00006421, "RTD1295", default_name, rtd1295_revisions, "Kylin" },
+	{ 0x00006421, "RTD1295", rtd1295_name, rtd1295_revisions, "Kylin" },
 };
 
Quoted 4 lines

static const struct rtd_soc *rtd_soc_by_chip_id(u32 chip_id)
--
2.16.4

_______________________________________________
linux-arm-kernel mailing list
[email protected]
http://lists.infradead.org/mailman/listinfo/linux-arm-kernel

#38 lore raw

RFC 7/11 arm64: dts: realtek: rtd129x: Extend chip-info reg with CHIP_INFO1

In reply to Andreas Färber

This additional register is needed to distinguish RTD1296 from RTD1295.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +2 -1
index 15a7c249155d..fea7c1ed7d08 100644
@@ -86,7 +86,8 @@
 
 		chip-info@9801a200 {
 			compatible = "realtek,rtd1195-chip";
-			reg = <0x9801a200 0x8>;
+			reg = <0x9801a200 0x8>,
+			      <0x98007028 0x4>;
 		};
 
 		uart1: serial@9801b200 {
Signature · 1 line
2.16.4
#39 lore raw

RFC 8/11 soc: realtek: chip: Detect RTD1293

In reply to Andreas Färber

Logic self-determined by comparing DS418j and DS418 registers.

---
 drivers/soc/realtek/chip.c | 2 ++
 1 file changed, 2 insertions(+)
drivers/soc/realtek/chip.c +2 -0
index ba653c097644..f4b26fb048c7 100644
@@ -59,6 +59,8 @@ static const char *rtd1295_name(struct device *dev, const struct rtd_soc *s)
 		u32 chipinfo1 = readl_relaxed(base);
 		iounmap(base);
 		if (chipinfo1 & BIT(11)) {
+			if (chipinfo1 & BIT(4))
+				return "RTD1293";
 			return "RTD1296";
 		}
 	}
Signature · 1 line
2.16.4
#40 lore raw

RFC 9/11 dt-bindings: soc: realtek: rtd1195-chip: Extend reg node again

In reply to Andreas Färber

Allow to optionally specify a third register to identify the chip. Whether needed and which register to specify depends on the family; RTD1295 family will want an efuse register.

---
I don't like specifying an efuse register here, which seems its own IP block.

 .../devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml    | 9 ++++++++-
 1 file changed, 8 insertions(+), 1 deletion(-)
Documentation/devicetree/bindings/soc/realtek/realtek,rtd1195-chip.yaml +8 -1
index e431cf559b66..249737e116d7 100644
@@ -19,7 +19,7 @@ properties:
 
   reg:
     minItems: 1
-    maxItems: 2
+    maxItems: 3
 
 required:
   - compatible
@@ -37,4 +37,11 @@ examples:
         reg = <0x9801a200 0x8>,
               <0x98007028 0x4>;
     };
+  - |
+    chip-info@9801a200 {
+        compatible = "realtek,rtd1195-chip";
+        reg = <0x9801a200 0x8>,
+              <0x98007028 0x4>,
+              <0x980171d8 0x4>;
+    };
 ...
Signature · 1 line
2.16.4
#41 lore raw

RFC 10/11 soc: realtek: chip: Detect RTD1294

In reply to Andreas Färber

Detection logic from downstream include/soc/realtek/rtd129x_cpu.h.

---
 drivers/soc/realtek/chip.c | 8 ++++++++
 1 file changed, 8 insertions(+)
drivers/soc/realtek/chip.c +8 -0
index f4b26fb048c7..e13339a4ca2e 100644
@@ -54,6 +54,14 @@ static const char *rtd1295_name(struct device *dev, const struct rtd_soc *s)
 {
 	void __iomem *base;
 
+	base = of_iomap(dev->of_node, 2);
+	if (base) {
+		u32 efuse = readl_relaxed(base);
+		iounmap(base);
+		if ((efuse & 0x3) == 0x1)
+			return "RTD1294";
+	}
+
 	base = of_iomap(dev->of_node, 1);
 	if (base) {
 		u32 chipinfo1 = readl_relaxed(base);
Signature · 1 line
2.16.4
#42 lore raw

RFC 11/11 arm64: dts: realtek: rtd129x: Extend chip-info reg with efuse

In reply to Andreas Färber

This register is needed to detect RTD1294.

---
 arch/arm64/boot/dts/realtek/rtd129x.dtsi | 3 ++-
 1 file changed, 2 insertions(+), 1 deletion(-)
arch/arm64/boot/dts/realtek/rtd129x.dtsi +2 -1
index fea7c1ed7d08..670efa86f661 100644
@@ -87,7 +87,8 @@
 		chip-info@9801a200 {
 			compatible = "realtek,rtd1195-chip";
 			reg = <0x9801a200 0x8>,
-			      <0x98007028 0x4>;
+			      <0x98007028 0x4>,
+			      <0x980171d8 0x4>;
 		};
 
 		uart1: serial@9801b200 {
Signature · 1 line
2.16.4
#43 lore raw
In reply to Andreas Färber

Am 03.11.19 um 02:36 schrieb Andreas Färber:

Quoted 7 lines

Prepared but not included here is:
* RTD1395 family, which we don't have a DT for yet,
* RTD1619 family, which we don't have a DT for yet, Chip ID to be verified,
* RTD1319 family, which we don't have a DT for yet, with TODO for its Chip ID.

Latest experimental patches at:
https://github.com/afaerber/linux/commits/rtd1295-next

For anyone wondering, the RTD1395 SoC info patch in above tree had a typo in the chip id that has been fixed and tested on BPi-M4 now.

Cheers,
Andreas

Signature · 4 lines
SUSE Software Solutions Germany GmbH
Maxfeldstr. 5, 90409 Nürnberg, Germany
GF: Felix Imendörffer
HRB 36809 (AG Nürnberg)