[PATCH v1 0/2] iio: core: Fix and cleanup to iio_device_alloc()

[PATCH v1 0/2] iio: core: Fix and cleanup to iio_device_alloc()

[Andy Shevchenko]

This is an exctract of the two patches from 
https://lore.kernel.org/r/20240228204919.3680786-1-andriy.shevchenko@linux.intel.com

In v1 (extract):
- removed unneeded 'else' branch (David)
- fixed typo in allocation cleanup (David)

Andy Shevchenko (2):
  iio: core: Leave private pointer NULL when no private data supplied
  iio: core: Calculate alloc_size only once in iio_device_alloc()

 drivers/iio/industrialio-core.c | 15 ++++++++-------
 1 file changed, 8 insertions(+), 7 deletions(-)

-- 
2.43.0.rc1.1.gbec44491f096

[PATCH v1 1/2] iio: core: Leave private pointer NULL when no private data supplied

[Andy Shevchenko]

In iio_device_alloc() when size of the private data is 0,
the private pointer is calculated to point behind the valid data.
Leave it NULL when no private data supplied.

Fixes: 6d4ebd565d15 ("iio: core: wrap IIO device into an iio_dev_opaque object")
Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/iio/industrialio-core.c | 6 ++++--
 1 file changed, 4 insertions(+), 2 deletions(-)

diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
index 4302093b92c7..8684ba246969 100644
--- a/drivers/iio/industrialio-core.c
+++ b/drivers/iio/industrialio-core.c
@@ -1654,8 +1654,10 @@ struct iio_dev *iio_device_alloc(struct device *parent, int sizeof_priv)
 		return NULL;
 
 	indio_dev = &iio_dev_opaque->indio_dev;
-	indio_dev->priv = (char *)iio_dev_opaque +
-		ALIGN(sizeof(struct iio_dev_opaque), IIO_DMA_MINALIGN);
+
+	if (sizeof_priv)
+		indio_dev->priv = (char *)iio_dev_opaque +
+			ALIGN(sizeof(*iio_dev_opaque), IIO_DMA_MINALIGN);
 
 	indio_dev->dev.parent = parent;
 	indio_dev->dev.type = &iio_device_type;
-- 
2.43.0.rc1.1.gbec44491f096

Re: [PATCH v1 1/2] iio: core: Leave private pointer NULL when no private data supplied

[David Lechner]

On Mon, Mar 4, 2024 at 8:07 AM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> In iio_device_alloc() when size of the private data is 0,
> the private pointer is calculated to point behind the valid data.
> Leave it NULL when no private data supplied.
>
> Fixes: 6d4ebd565d15 ("iio: core: wrap IIO device into an iio_dev_opaque object")
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---

Reviewed-by: David Lechner <dlechner@baylibre.com>

Re: [PATCH v1 1/2] iio: core: Leave private pointer NULL when no private data supplied

[Jonathan Cameron]

On Mon,  4 Mar 2024 16:04:32 +0200
Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:

> In iio_device_alloc() when size of the private data is 0,
> the private pointer is calculated to point behind the valid data.
> Leave it NULL when no private data supplied.
> 
> Fixes: 6d4ebd565d15 ("iio: core: wrap IIO device into an iio_dev_opaque object")
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>

Fix is a strong word given any driver trying to access the pointer in this case
will be broken anyway.  I don't mind it being backported as it'll make bugs
easier to identify but I'm not rushing it in.

Otherwise seems like a sensible bit of tidying up.

Jonathan

> ---
>  drivers/iio/industrialio-core.c | 6 ++++--
>  1 file changed, 4 insertions(+), 2 deletions(-)
> 
> diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
> index 4302093b92c7..8684ba246969 100644
> --- a/drivers/iio/industrialio-core.c
> +++ b/drivers/iio/industrialio-core.c
> @@ -1654,8 +1654,10 @@ struct iio_dev *iio_device_alloc(struct device *parent, int sizeof_priv)
>  		return NULL;
>  
>  	indio_dev = &iio_dev_opaque->indio_dev;
> -	indio_dev->priv = (char *)iio_dev_opaque +
> -		ALIGN(sizeof(struct iio_dev_opaque), IIO_DMA_MINALIGN);
> +
> +	if (sizeof_priv)
> +		indio_dev->priv = (char *)iio_dev_opaque +
> +			ALIGN(sizeof(*iio_dev_opaque), IIO_DMA_MINALIGN);
>  
>  	indio_dev->dev.parent = parent;
>  	indio_dev->dev.type = &iio_device_type;

[PATCH v1 2/2] iio: core: Calculate alloc_size only once in iio_device_alloc()

[Andy Shevchenko]

No need to rewrite the value, instead use 'else' branch.
This will also help further refactoring the code later on.

Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
---
 drivers/iio/industrialio-core.c | 9 ++++-----
 1 file changed, 4 insertions(+), 5 deletions(-)

diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
index 8684ba246969..c7ad88932015 100644
--- a/drivers/iio/industrialio-core.c
+++ b/drivers/iio/industrialio-core.c
@@ -1643,11 +1643,10 @@ struct iio_dev *iio_device_alloc(struct device *parent, int sizeof_priv)
 	struct iio_dev *indio_dev;
 	size_t alloc_size;
 
-	alloc_size = sizeof(struct iio_dev_opaque);
-	if (sizeof_priv) {
-		alloc_size = ALIGN(alloc_size, IIO_DMA_MINALIGN);
-		alloc_size += sizeof_priv;
-	}
+	if (sizeof_priv)
+		alloc_size = ALIGN(sizeof(*iio_dev_opaque), IIO_DMA_MINALIGN) + sizeof_priv;
+	else
+		alloc_size = sizeof(*iio_dev_opaque);
 
 	iio_dev_opaque = kzalloc(alloc_size, GFP_KERNEL);
 	if (!iio_dev_opaque)
-- 
2.43.0.rc1.1.gbec44491f096

Re: [PATCH v1 2/2] iio: core: Calculate alloc_size only once in iio_device_alloc()

[David Lechner]

On Mon, Mar 4, 2024 at 8:07 AM Andy Shevchenko
<andriy.shevchenko@linux.intel.com> wrote:
>
> No need to rewrite the value, instead use 'else' branch.
> This will also help further refactoring the code later on.
>
> Signed-off-by: Andy Shevchenko <andriy.shevchenko@linux.intel.com>
> ---

Reviewed-by: David Lechner <dlechner@baylibre.com>

Re: [PATCH v1 2/2] iio: core: Calculate alloc_size only once in iio_device_alloc()

[Christophe JAILLET]

Le 04/03/2024 à 15:04, Andy Shevchenko a écrit :
> No need to rewrite the value, instead use 'else' branch.
> This will also help further refactoring the code later on.
> 
> Signed-off-by: Andy Shevchenko <andriy.shevchenko-VuQAYsv1563Yd54FQh9/CA@public.gmane.org>
> ---
>   drivers/iio/industrialio-core.c | 9 ++++-----
>   1 file changed, 4 insertions(+), 5 deletions(-)
> 
> diff --git a/drivers/iio/industrialio-core.c b/drivers/iio/industrialio-core.c
> index 8684ba246969..c7ad88932015 100644
> --- a/drivers/iio/industrialio-core.c
> +++ b/drivers/iio/industrialio-core.c
> @@ -1643,11 +1643,10 @@ struct iio_dev *iio_device_alloc(struct device *parent, int sizeof_priv)
>   	struct iio_dev *indio_dev;
>   	size_t alloc_size;
>   
> -	alloc_size = sizeof(struct iio_dev_opaque);
> -	if (sizeof_priv) {
> -		alloc_size = ALIGN(alloc_size, IIO_DMA_MINALIGN);
> -		alloc_size += sizeof_priv;
> -	}
> +	if (sizeof_priv)
> +		alloc_size = ALIGN(sizeof(*iio_dev_opaque), IIO_DMA_MINALIGN) + sizeof_priv;

maybe size_add() to be super-safe?

CJ

> +	else
> +		alloc_size = sizeof(*iio_dev_opaque);
>   
>   	iio_dev_opaque = kzalloc(alloc_size, GFP_KERNEL);
>   	if (!iio_dev_opaque)

Re: [PATCH v1 0/2] iio: core: Fix and cleanup to iio_device_alloc()

[Jonathan Cameron]

On Mon,  4 Mar 2024 16:04:31 +0200
Andy Shevchenko <andriy.shevchenko@linux.intel.com> wrote:

> This is an exctract of the two patches from 
> https://lore.kernel.org/r/20240228204919.3680786-1-andriy.shevchenko@linux.intel.com
> 
> In v1 (extract):
> - removed unneeded 'else' branch (David)
> - fixed typo in allocation cleanup (David)
> 
> Andy Shevchenko (2):
>   iio: core: Leave private pointer NULL when no private data supplied
>   iio: core: Calculate alloc_size only once in iio_device_alloc()
> 
>  drivers/iio/industrialio-core.c | 15 ++++++++-------
>  1 file changed, 8 insertions(+), 7 deletions(-)
> 

Applied to the togreg-normal branch of iio.git which I'll rebase on rc1
once available and push out as togreg.

Thanks,

Jonathan