Skip to contentSkip to conversation

staging: iio: ad2s1210: Destroy mutex at remove

PATCH In review Tallys Martins 3 participants 5 messages May 18, 2019 – May 20, 2019

Subsystems (4)
lore
Share & tools
lore
#0 lore raw

PATCH 1/2 staging: iio: ad2s1210: Destroy mutex at remove

Ensure the mutex will be destroyed on drive removal. Also adds mutex comment description.

---
 drivers/staging/iio/resolver/ad2s1210.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/staging/iio/resolver/ad2s1210.c +3 -1
index b6be0bc202f5..b91cf57c5e57 100644
@@ -86,7 +86,7 @@ static const struct ad2s1210_gpio gpios[] = {
 static const unsigned int ad2s1210_resolution_value[] = { 10, 12, 14, 16 };
 
 struct ad2s1210_state {
-	struct mutex lock;
+	struct mutex lock; /* lock to protect the state on r/w operations */
 	struct spi_device *sdev;
 	struct gpio_desc *gpios[5];
 	unsigned int fclkin;
@@ -689,8 +689,10 @@ static int ad2s1210_probe(struct spi_device *spi)
 static int ad2s1210_remove(struct spi_device *spi)
 {
 	struct iio_dev *indio_dev = spi_get_drvdata(spi);
+	struct ad2s1210_state *ad2s1210_ad = iio_priv(indio_dev);
 
 	iio_device_unregister(indio_dev);
+	mutex_destroy(&ad2s1210_ad->lock);
 
 	return 0;
 }
Signature · 1 line
2.21.0
#1 lore raw

PATCH 2/2 staging: iio: ad2s1210: Add devicetree yaml doc

The driver currently has no devicetree documentation. This commit adds a devicetree folder and documentation for it. Documentation must be moved as well when the driver gets out of staging.

---
 .../Documentation/devicetree/ad2s1210.yaml    | 41 +++++++++++++++++++
 1 file changed, 41 insertions(+)
 create mode 100644 drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +41 -0
new file mode 100644
index 000000000000..733aa07b4626
@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: GPL-2.0
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/iio/iio/ad2s1210.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: |
+  Analog Devices Inc. AD2S1210 10-Bit to 16-Bit R/D Converters
+
+maintainers:
+  - Graff Yang <[email protected]>
+
+description: |
+  Analog Devices AD2S1210 Resolver to Digital SPI driver
+
+  https://www.analog.com/en/products/ad2s1210.html
+
+properties:
+  compatible:
+    enum:
+      - adi,ad2s1210
+
+  reg:
+    maxItems: 1
+
+  clock-frequency:
+    minimum: 2000
+    maximum: 20000
+    default: 8192
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+  resolver@0 {
+    compatible = "adi,ad2s1210";
+    reg = <0>;
+  };
+...
Signature · 1 line
2.21.0
maintainer
#2 lore raw

Re: [PATCH 2/2] staging: iio: ad2s1210: Add devicetree yaml doc

Quoted 7 lines · On Sat, 18 May 2019 19:15:58 -0300 Tallys Martins <[email protected]>…

The driver currently has no devicetree documentation. This commit adds a devicetree folder and documentation for it. Documentation must be moved as well when the driver gets out of staging.

Hi,

There is no need for a direct relationship between a binding and a driver at all. As such, we normally don't take binding documents in staging.

Just put it directly in it's final destination. The driver can catch up with it later!

I'm still not that comfortable with yaml (haven't gotten around to doing any conversions myself yet) so I'll be looking for additional review on these from others.

A few comments inline.

Quoted 12 lines
---
 .../Documentation/devicetree/ad2s1210.yaml    | 41 +++++++++++++++++++
 1 file changed, 41 insertions(+)
 create mode 100644 drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +1 -0
new file mode 100644
index 000000000000..733aa07b4626
@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: GPL-2.0

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 11

Quoted 10 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +10 -0
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/iio/iio/ad2s1210.yaml#
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: |
+  Analog Devices Inc. AD2S1210 10-Bit to 16-Bit R/D Converters
+
+maintainers:
+  - Graff Yang <[email protected]>

I would check that one with the Analog devices team.

Quoted 8 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +8 -0
+
+description: |
+  Analog Devices AD2S1210 Resolver to Digital SPI driver
+
+  https://www.analog.com/en/products/ad2s1210.html
+
+properties:
+  compatible:

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 29

Quoted 10 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +10 -0
+    enum:
+      - adi,ad2s1210
+
+  reg:
+    maxItems: 1
+
+  clock-frequency:
+    minimum: 2000
+    maximum: 20000
+    default: 8192

This doesn't look like a modern clock binding.
If we are going to end up changing this, then we should probably delay having a binding doc until after that change is made.

We often only do binding docs for drivers in staging just before moving them out so as to avoid this sort of issue.

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 39

Quoted 10 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +10 -0
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+  resolver@0 {
+    compatible = "adi,ad2s1210";
+    reg = <0>;

An example is probably more useful if it includes all the optional properties as well.

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 41

Quoted 2 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +2 -0
+  };
+...

Thanks,

Jonathan

#3 lore raw

Re: [PATCH 2/2] staging: iio: ad2s1210: Add devicetree yaml doc

Quoted 4 lines · Jonathan Cameron

On Sat, 18 May 2019 19:15:58 -0300
Tallys Martins <[email protected]> wrote:

CC-ing my work-email
There are some issues with it and mailing lists; I'll hopefully sort them out in the next weeks.

Quoted 13 lines
Quoted 7 lines

The driver currently has no devicetree documentation. This commit adds a devicetree folder and documentation for it. Documentation must be moved as well when the driver gets out of staging.

Hi,

There is no need for a direct relationship between a binding and a driver at all. As such, we normally don't take binding documents in staging.

Quoted 6 lines

Just put it directly in it's final destination. The driver can catch up with it later!

I'm still not that comfortable with yaml (haven't gotten around to doing any conversions myself yet) so I'll be looking for additional review on these from others.

Personal note: I also don't like YAML, but that may be me feeling old. I'm not that old to prefer XML, but old enough to prefer JSON. I am still trying to accommodate my taste/feeling for the format, even though I've been using it sporadically for ~5 years now in various other elements.
Travis-CI may be the biggest user of it.
Looks like Ruby users were the biggest pushers/initiators of YAML, and Python people seem to have jumped in shortly after.

Oh well: ¯\_(ツ)_/¯

Quoted 7 lines

A few comments inline.

Quoted 4 lines
---
 .../Documentation/devicetree/ad2s1210.yaml    | 41 +++++++++++++++++++
 1 file changed, 41 insertions(+)
 create mode 100644 drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml

Maybe also remove the old txt binding if adding this new one. No need to keep them both.

Quoted 11 lines
Quoted 11 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +4 -0
new file mode 100644
index 000000000000..733aa07b4626
@@ -0,0 +1,41 @@
+# SPDX-License-Identifier: GPL-2.0
+%YAML 1.2
+---
+$id: http://devicetree.org/schemas/iio/iio/ad2s1210.yaml#
Quoted 8 lines

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 11

Quoted 7 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +7 -0
+$schema: http://devicetree.org/meta-schemas/core.yaml#
+
+title: |
+  Analog Devices Inc. AD2S1210 10-Bit to 16-Bit R/D Converters
+
+maintainers:
+  - Graff Yang <[email protected]>

I would check that one with the Analog devices team.

I guess, add (here):
Michael Hennerich <[email protected]>

Quoted 11 lines

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 21

Quoted 11 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +10 -0
+
+description: |
+  Analog Devices AD2S1210 Resolver to Digital SPI driver
+
+  https://www.analog.com/en/products/ad2s1210.html
+
+properties:
+  compatible:
+    enum:
+      - adi,ad2s1210

These 2 are required here, since this chip operates in SPI mode 3.

spi-cpha: true

spi-cpol: true

Also, from the driver, some GPIOs should also be documented:

static const struct ad2s1210_gpio gpios[] = {
        [AD2S1210_SAMPLE] = { .name = "adi,sample", .flags = GPIOD_OUT_LOW },
        [AD2S1210_A0] = { .name = "adi,a0", .flags = GPIOD_OUT_LOW },
        [AD2S1210_A1] = { .name = "adi,a1", .flags = GPIOD_OUT_LOW },
        [AD2S1210_RES0] = { .name = "adi,res0", .flags = GPIOD_OUT_LOW },
        [AD2S1210_RES1] = { .name = "adi,res1", .flags = GPIOD_OUT_LOW },
};

So,

adi,sample-gpios:
adi,a0-gpios:
adi,a1-gpios:
adi,res0-gpios:
adi,res10-gpios:

These also look like they are required.

Quoted 12 lines

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 29

Quoted 8 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +8 -0
+
+  reg:
+    maxItems: 1
+
+  clock-frequency:
+    minimum: 2000
+    maximum: 20000
+    default: 8192

This doesn't look like a modern clock binding.
If we are going to end up changing this, then we should probably delay having a binding doc until after that change is made.

Quoted 2 lines

We often only do binding docs for drivers in staging just before moving them out so as to avoid this sort of issue.

These clock properties are not needed here.

Quoted 13 lines

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 39

Quoted 11 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +10 -0
+
+required:
+  - compatible
+  - reg
+
+examples:
+  - |
+  resolver@0 {
+    compatible = "adi,ad2s1210";
+    reg = <0>;

An example is probably more useful if it includes all the optional properties as well.

This should also be included in an spi block:
So:

spi0 {
       resolver@0 {
            compatible = "adi,ad2s1210";
            reg = <0>;
            spi-cpha;
            spi-cpol;
            // and then the adi,***-gpios I think here
       };
};

I don't think much more-else is needed for this driver.

Quoted 7 lines

drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml line 41

Quoted 2 lines
drivers/staging/iio/Documentation/devicetree/ad2s1210.yaml new file +2 -0
+  };
+...

Thanks,

Jonathan

maintainer
#4 lore raw
In reply to Tallys Martins
Quoted 6 lines · On Sat, 18 May 2019 19:15:57 -0300 Tallys Martins <[email protected]>…

Ensure the mutex will be destroyed on drive removal. Also adds mutex comment description.

Hi.

This particular 'issue' is never a simple one.

The destroy_mutex call is intended for lock debugging only, which is why there is devm_init_mutex or similar to allow for automatic unwinding.

It simple cleanup paths like this, it provides very little value and leads to a more complex unwind.

It is for this reason that the vast majority of drivers simply don't bother.

Hence, unless there is a good reason I'm missing I won't be introducing more of them.

Thanks,

Jonathan

Quoted 28 lines
---
 drivers/staging/iio/resolver/ad2s1210.c | 4 +++-
 1 file changed, 3 insertions(+), 1 deletion(-)
drivers/staging/iio/resolver/ad2s1210.c +3 -1
index b6be0bc202f5..b91cf57c5e57 100644
@@ -86,7 +86,7 @@ static const struct ad2s1210_gpio gpios[] = {
 static const unsigned int ad2s1210_resolution_value[] = { 10, 12, 14, 16 };
 
 struct ad2s1210_state {
-	struct mutex lock;
+	struct mutex lock; /* lock to protect the state on r/w operations */
 	struct spi_device *sdev;
 	struct gpio_desc *gpios[5];
 	unsigned int fclkin;
@@ -689,8 +689,10 @@ static int ad2s1210_probe(struct spi_device *spi)
 static int ad2s1210_remove(struct spi_device *spi)
 {
 	struct iio_dev *indio_dev = spi_get_drvdata(spi);
+	struct ad2s1210_state *ad2s1210_ad = iio_priv(indio_dev);
 
 	iio_device_unregister(indio_dev);
+	mutex_destroy(&ad2s1210_ad->lock);
 
 	return 0;
 }