[RFC PATCH v4 0/2] Small devm helper for devm implementations

[RFC PATCH v4 0/2] Small devm helper for devm implementations

[Marc Gonzalez]

Hello,

Differences from v3 to v4
x Add a bunch of kerneldoc above devm_add() [Greg KH]
x Split patch in two [Greg KH]

Differences from v2 to v3
x Make devm_add() return an error-code rather than the raw data pointer
  (in case devres_alloc ever returns an ERR_PTR) as suggested by Geert
x Provide a variadic version devm_vadd() to work with structs as suggested
  by Geert
x Don't use nested ifs in clk_devm* implementations (hopefully simpler
  code logic to follow) as suggested by Geert

Marc Gonzalez (2):
  devres: Provide new helper for devm functions
  clk: Use devm_add in managed functions

 drivers/base/devres.c    |  28 +++++++++++
 drivers/clk/clk-devres.c | 101 ++++++++++++++-------------------------
 include/linux/device.h   |   3 ++
 3 files changed, 68 insertions(+), 64 deletions(-)

-- 
2.17.1

[RFC PATCH v4 1/2] devres: Provide new helper for devm functions

[Marc Gonzalez]

Provide a simple wrapper for devres_alloc / devres_add.

Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr>
---
 drivers/base/devres.c  | 28 ++++++++++++++++++++++++++++
 include/linux/device.h |  3 +++
 2 files changed, 31 insertions(+)

diff --git a/drivers/base/devres.c b/drivers/base/devres.c
index 0bbb328bd17f..7fe6cc34411e 100644
--- a/drivers/base/devres.c
+++ b/drivers/base/devres.c
@@ -685,6 +685,34 @@ int devres_release_group(struct device *dev, void *id)
 }
 EXPORT_SYMBOL_GPL(devres_release_group);
 
+/**
+ * devm_add - allocate and register new device resource
+ * @dev: device to add resource to
+ * @func: resource release function
+ * @arg: resource data
+ * @size: resource data size
+ *
+ * Simple wrapper for devres_alloc / devres_add.
+ * Release the resource if the allocation fails.
+ *
+ * RETURNS:
+ * 0 if the allocation succeeds, -ENOMEM otherwise.
+ */
+int devm_add(struct device *dev, dr_release_t func, void *arg, size_t size)
+{
+	void *data = devres_alloc(func, size, GFP_KERNEL);
+
+	if (!data) {
+		func(dev, arg);
+		return -ENOMEM;
+	}
+
+	memcpy(data, arg, size);
+	devres_add(dev, data);
+	return 0;
+}
+EXPORT_SYMBOL_GPL(devm_add);
+
 /*
  * Custom devres actions allow inserting a simple function call
  * into the teadown sequence.
diff --git a/include/linux/device.h b/include/linux/device.h
index 0cd7c647c16c..55be3be9b276 100644
--- a/include/linux/device.h
+++ b/include/linux/device.h
@@ -247,6 +247,9 @@ void __iomem *devm_of_iomap(struct device *dev,
 			    struct device_node *node, int index,
 			    resource_size_t *size);
 
+int devm_add(struct device *dev, dr_release_t func, void *arg, size_t size);
+#define devm_vadd(dev, func, type, args...) \
+	devm_add(dev, func, &(struct type){args}, sizeof(struct type))
 /* allows to add/remove a custom action to devres stack */
 int devm_add_action(struct device *dev, void (*action)(void *), void *data);
 void devm_remove_action(struct device *dev, void (*action)(void *), void *data);
-- 
2.17.1

Re: [RFC PATCH v4 1/2] devres: Provide new helper for devm functions

[Geert Uytterhoeven]

Hi Marc,

On Wed, Feb 26, 2020 at 4:55 PM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
> Provide a simple wrapper for devres_alloc / devres_add.
>
> Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr>

With the grammar fixed, as per below:
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

> --- a/drivers/base/devres.c
> +++ b/drivers/base/devres.c
> @@ -685,6 +685,34 @@ int devres_release_group(struct device *dev, void *id)
>  }
>  EXPORT_SYMBOL_GPL(devres_release_group);
>
> +/**
> + * devm_add - allocate and register new device resource
> + * @dev: device to add resource to
> + * @func: resource release function
> + * @arg: resource data
> + * @size: resource data size
> + *
> + * Simple wrapper for devres_alloc / devres_add.
> + * Release the resource if the allocation fails.

Releases ... failed.

> + *
> + * RETURNS:
> + * 0 if the allocation succeeds, -ENOMEM otherwise.

"0 on success" would avoid any discussion about "succeeds" or "succeeded" ;-)

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

[RFC PATCH v4 2/2] clk: Use devm_add in managed functions

[Marc Gonzalez]

Using the helper produces simpler code, and smaller object size.
E.g. with gcc-arm-9.2-2019.12-x86_64-aarch64-none-linux-gnu:

    text           data     bss     dec     hex filename
-   1708             80       0    1788     6fc drivers/clk/clk-devres.o
+   1524             80       0    1604     644 drivers/clk/clk-devres.o

Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr>
---
 drivers/clk/clk-devres.c | 101 ++++++++++++++-------------------------
 1 file changed, 37 insertions(+), 64 deletions(-)

diff --git a/drivers/clk/clk-devres.c b/drivers/clk/clk-devres.c
index be160764911b..3f4b200b5328 100644
--- a/drivers/clk/clk-devres.c
+++ b/drivers/clk/clk-devres.c
@@ -4,26 +4,22 @@
 #include <linux/export.h>
 #include <linux/gfp.h>
 
-static void devm_clk_release(struct device *dev, void *res)
+static void my_clk_put(struct device *dev, void *res)
 {
 	clk_put(*(struct clk **)res);
 }
 
 struct clk *devm_clk_get(struct device *dev, const char *id)
 {
-	struct clk **ptr, *clk;
-
-	ptr = devres_alloc(devm_clk_release, sizeof(*ptr), GFP_KERNEL);
-	if (!ptr)
-		return ERR_PTR(-ENOMEM);
-
-	clk = clk_get(dev, id);
-	if (!IS_ERR(clk)) {
-		*ptr = clk;
-		devres_add(dev, ptr);
-	} else {
-		devres_free(ptr);
-	}
+	int ret;
+	struct clk *clk = clk_get(dev, id);
+
+	if (IS_ERR(clk))
+		return clk;
+
+	ret = devm_add(dev, my_clk_put, &clk, sizeof(clk));
+	if (ret)
+		return ERR_PTR(ret);
 
 	return clk;
 }
@@ -40,14 +36,14 @@ struct clk *devm_clk_get_optional(struct device *dev, const char *id)
 }
 EXPORT_SYMBOL(devm_clk_get_optional);
 
-struct clk_bulk_devres {
-	struct clk_bulk_data *clks;
+struct clk_bulk_args {
 	int num_clks;
+	struct clk_bulk_data *clks;
 };
 
-static void devm_clk_bulk_release(struct device *dev, void *res)
+static void my_clk_bulk_put(struct device *dev, void *res)
 {
-	struct clk_bulk_devres *devres = res;
+	struct clk_bulk_args *devres = res;
 
 	clk_bulk_put(devres->num_clks, devres->clks);
 }
@@ -55,25 +51,17 @@ static void devm_clk_bulk_release(struct device *dev, void *res)
 static int __devm_clk_bulk_get(struct device *dev, int num_clks,
 			       struct clk_bulk_data *clks, bool optional)
 {
-	struct clk_bulk_devres *devres;
 	int ret;
 
-	devres = devres_alloc(devm_clk_bulk_release,
-			      sizeof(*devres), GFP_KERNEL);
-	if (!devres)
-		return -ENOMEM;
-
 	if (optional)
 		ret = clk_bulk_get_optional(dev, num_clks, clks);
 	else
 		ret = clk_bulk_get(dev, num_clks, clks);
-	if (!ret) {
-		devres->clks = clks;
-		devres->num_clks = num_clks;
-		devres_add(dev, devres);
-	} else {
-		devres_free(devres);
-	}
+
+	if (ret)
+		return ret;
+
+	ret = devm_vadd(dev, my_clk_bulk_put, clk_bulk_args, num_clks, clks);
 
 	return ret;
 }
@@ -95,24 +83,17 @@ EXPORT_SYMBOL_GPL(devm_clk_bulk_get_optional);
 int __must_check devm_clk_bulk_get_all(struct device *dev,
 				       struct clk_bulk_data **clks)
 {
-	struct clk_bulk_devres *devres;
 	int ret;
+	int num_clks = clk_bulk_get_all(dev, clks);
 
-	devres = devres_alloc(devm_clk_bulk_release,
-			      sizeof(*devres), GFP_KERNEL);
-	if (!devres)
-		return -ENOMEM;
-
-	ret = clk_bulk_get_all(dev, &devres->clks);
-	if (ret > 0) {
-		*clks = devres->clks;
-		devres->num_clks = ret;
-		devres_add(dev, devres);
-	} else {
-		devres_free(devres);
-	}
+	if (num_clks <= 0)
+		return num_clks;
 
-	return ret;
+	ret = devm_vadd(dev, my_clk_bulk_put, clk_bulk_args, num_clks, *clks);
+	if (ret)
+		return ret;
+
+	return num_clks;
 }
 EXPORT_SYMBOL_GPL(devm_clk_bulk_get_all);
 
@@ -128,30 +109,22 @@ static int devm_clk_match(struct device *dev, void *res, void *data)
 
 void devm_clk_put(struct device *dev, struct clk *clk)
 {
-	int ret;
-
-	ret = devres_release(dev, devm_clk_release, devm_clk_match, clk);
-
-	WARN_ON(ret);
+	WARN_ON(devres_release(dev, my_clk_put, devm_clk_match, clk));
 }
 EXPORT_SYMBOL(devm_clk_put);
 
 struct clk *devm_get_clk_from_child(struct device *dev,
 				    struct device_node *np, const char *con_id)
 {
-	struct clk **ptr, *clk;
-
-	ptr = devres_alloc(devm_clk_release, sizeof(*ptr), GFP_KERNEL);
-	if (!ptr)
-		return ERR_PTR(-ENOMEM);
-
-	clk = of_clk_get_by_name(np, con_id);
-	if (!IS_ERR(clk)) {
-		*ptr = clk;
-		devres_add(dev, ptr);
-	} else {
-		devres_free(ptr);
-	}
+	int ret;
+	struct clk *clk = of_clk_get_by_name(np, con_id);
+
+	if (IS_ERR(clk))
+		return clk;
+
+	ret = devm_add(dev, my_clk_put, &clk, sizeof(clk));
+	if (ret)
+		return ERR_PTR(ret);
 
 	return clk;
 }
-- 
2.17.1

Re: [RFC PATCH v4 2/2] clk: Use devm_add in managed functions

[Geert Uytterhoeven]

Hi Marc,

Thanks for your patch!

On Wed, Feb 26, 2020 at 4:55 PM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
> Using the helper produces simpler code, and smaller object size.
> E.g. with gcc-arm-9.2-2019.12-x86_64-aarch64-none-linux-gnu:
>
>     text           data     bss     dec     hex filename
> -   1708             80       0    1788     6fc drivers/clk/clk-devres.o
> +   1524             80       0    1604     644 drivers/clk/clk-devres.o

And the size reduction could have been even more ;-)

> Signed-off-by: Marc Gonzalez <marc.w.gonzalez@free.fr>

> --- a/drivers/clk/clk-devres.c
> +++ b/drivers/clk/clk-devres.c

> @@ -55,25 +51,17 @@ static void devm_clk_bulk_release(struct device *dev, void *res)
>  static int __devm_clk_bulk_get(struct device *dev, int num_clks,
>                                struct clk_bulk_data *clks, bool optional)
>  {
> -       struct clk_bulk_devres *devres;
>         int ret;
>
> -       devres = devres_alloc(devm_clk_bulk_release,
> -                             sizeof(*devres), GFP_KERNEL);
> -       if (!devres)
> -               return -ENOMEM;
> -
>         if (optional)
>                 ret = clk_bulk_get_optional(dev, num_clks, clks);
>         else
>                 ret = clk_bulk_get(dev, num_clks, clks);
> -       if (!ret) {
> -               devres->clks = clks;
> -               devres->num_clks = num_clks;
> -               devres_add(dev, devres);
> -       } else {
> -               devres_free(devres);
> -       }
> +
> +       if (ret)
> +               return ret;
> +
> +       ret = devm_vadd(dev, my_clk_bulk_put, clk_bulk_args, num_clks, clks);
>
>         return ret;

return devm_vadd(...);

>  }

> @@ -128,30 +109,22 @@ static int devm_clk_match(struct device *dev, void *res, void *data)
>
>  void devm_clk_put(struct device *dev, struct clk *clk)
>  {
> -       int ret;
> -
> -       ret = devres_release(dev, devm_clk_release, devm_clk_match, clk);
> -
> -       WARN_ON(ret);
> +       WARN_ON(devres_release(dev, my_clk_put, devm_clk_match, clk));

Getting rid of "ret" is an unrelated change, which actually increases
kernel size, as the WARN_ON() parameter is stringified for the warning
message.

The rest looks good, so with the above fixed:
Reviewed-by: Geert Uytterhoeven <geert+renesas@glider.be>

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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

Re: [RFC PATCH v4 2/2] clk: Use devm_add in managed functions

[Marc Gonzalez]

On 27/02/2020 14:36, Geert Uytterhoeven wrote:

> Hi Marc,
> 
> Thanks for your patch!
> 
> On Wed, Feb 26, 2020 at 4:55 PM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
>> Using the helper produces simpler code, and smaller object size.
>> E.g. with gcc-arm-9.2-2019.12-x86_64-aarch64-none-linux-gnu:
>>
>>     text           data     bss     dec     hex filename
>> -   1708             80       0    1788     6fc drivers/clk/clk-devres.o
>> +   1524             80       0    1604     644 drivers/clk/clk-devres.o
> 
> And the size reduction could have been even more ;-)

I'll see what I can do! ;-)

I have another patch with even smaller object code, but it requires
C11 to be well-defined (memcmp the whole struct, which requires zeros
in the padding holes).


>> --- a/drivers/clk/clk-devres.c
>> +++ b/drivers/clk/clk-devres.c
> 
>> @@ -55,25 +51,17 @@ static void devm_clk_bulk_release(struct device *dev, void *res)
>>  static int __devm_clk_bulk_get(struct device *dev, int num_clks,
>>                                struct clk_bulk_data *clks, bool optional)
>>  {
>> -       struct clk_bulk_devres *devres;
>>         int ret;
>>
>> -       devres = devres_alloc(devm_clk_bulk_release,
>> -                             sizeof(*devres), GFP_KERNEL);
>> -       if (!devres)
>> -               return -ENOMEM;
>> -
>>         if (optional)
>>                 ret = clk_bulk_get_optional(dev, num_clks, clks);
>>         else
>>                 ret = clk_bulk_get(dev, num_clks, clks);
>> -       if (!ret) {
>> -               devres->clks = clks;
>> -               devres->num_clks = num_clks;
>> -               devres_add(dev, devres);
>> -       } else {
>> -               devres_free(devres);
>> -       }
>> +
>> +       if (ret)
>> +               return ret;
>> +
>> +       ret = devm_vadd(dev, my_clk_bulk_put, clk_bulk_args, num_clks, clks);
>>
>>         return ret;
> 
> return devm_vadd(...);

If you think that makes it look better, I'll make the change!


>> @@ -128,30 +109,22 @@ static int devm_clk_match(struct device *dev, void *res, void *data)
>>
>>  void devm_clk_put(struct device *dev, struct clk *clk)
>>  {
>> -       int ret;
>> -
>> -       ret = devres_release(dev, devm_clk_release, devm_clk_match, clk);
>> -
>> -       WARN_ON(ret);
>> +       WARN_ON(devres_release(dev, my_clk_put, devm_clk_match, clk));
> 
> Getting rid of "ret" is an unrelated change, which actually increases
> kernel size, as the WARN_ON() parameter is stringified for the warning
> message.

Weird... Are you sure about that? I built the preprocessed file,
and it didn't appear to be so.

#ifndef WARN_ON
#define WARN_ON(condition) ({						\
	int __ret_warn_on = !!(condition);				\
	if (unlikely(__ret_warn_on))					\
		__WARN();						\
	unlikely(__ret_warn_on);					\
})
#endif

Maybe you were thinking of i915's WARN_ON?

#define WARN_ON(x) WARN((x), "%s", "WARN_ON(" __stringify(x) ")")

Regards.

Re: [RFC PATCH v4 2/2] clk: Use devm_add in managed functions

[Geert Uytterhoeven]

Hi Marc,

On Mon, Mar 2, 2020 at 11:01 AM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
> On 27/02/2020 14:36, Geert Uytterhoeven wrote:
> > On Wed, Feb 26, 2020 at 4:55 PM Marc Gonzalez <marc.w.gonzalez@free.fr> wrote:
> >> Using the helper produces simpler code, and smaller object size.

> >> --- a/drivers/clk/clk-devres.c
> >> +++ b/drivers/clk/clk-devres.c

> >> @@ -128,30 +109,22 @@ static int devm_clk_match(struct device *dev, void *res, void *data)
> >>
> >>  void devm_clk_put(struct device *dev, struct clk *clk)
> >>  {
> >> -       int ret;
> >> -
> >> -       ret = devres_release(dev, devm_clk_release, devm_clk_match, clk);
> >> -
> >> -       WARN_ON(ret);
> >> +       WARN_ON(devres_release(dev, my_clk_put, devm_clk_match, clk));
> >
> > Getting rid of "ret" is an unrelated change, which actually increases
> > kernel size, as the WARN_ON() parameter is stringified for the warning
> > message.
>
> Weird... Are you sure about that? I built the preprocessed file,
> and it didn't appear to be so.
>
> #ifndef WARN_ON
> #define WARN_ON(condition) ({                                           \
>         int __ret_warn_on = !!(condition);                              \
>         if (unlikely(__ret_warn_on))                                    \
>                 __WARN();                                               \
>         unlikely(__ret_warn_on);                                        \
> })
> #endif
>
> Maybe you were thinking of i915's WARN_ON?
>
> #define WARN_ON(x) WARN((x), "%s", "WARN_ON(" __stringify(x) ")")

Oops, you're right.  I got trapped again by an override of a standard macro
(IMHO this should be removed).

Gr{oetje,eeting}s,

                        Geert

-- 
Geert Uytterhoeven -- There's lots of Linux beyond ia32 -- geert@linux-m68k.org

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