[PATCH] media: atomisp: use kmalloc_objs for array allocations

[PATCH] media: atomisp: use kmalloc_objs for array allocations

[Pedro Pontes]

Convert manual kmalloc() multiplications to the modern kmalloc_objs()
interface to improve type safety and prevent potential integer
overflows.

Signed-off-by: Pedro Pontes <pontescpedro@gmail.com>
---
This patch is a refresh and modernization of the following work by
Qianfeng Rong from August 2025:
Link: https://lore.kernel.org/all/20250821081746.528018-1-rongqianfeng@vivo.com/

The original version used manual multiplications; I have updated it
to use the newer kmalloc_objs() macro and enforced pointer-based
type inference (e.g., *descr->in_info) as is now the preferred
standard for these cleanups.

The patch has been verified with a W=1 build for pci/sh_css.o. Since I
don't own the hardware.

Please note that, since I'm a new contributor, I wasn't quite sure in
how to proceed in this scenario of "redoing" an older contribution.
I've credited the original author with the "Originally-authored-by"
tag. Please let me know if I should have done something differently.

 drivers/staging/media/atomisp/pci/sh_css.c | 57 ++++++++++++----------
 1 file changed, 30 insertions(+), 27 deletions(-)

diff --git a/drivers/staging/media/atomisp/pci/sh_css.c b/drivers/staging/media/atomisp/pci/sh_css.c
index 6cda5925fa45..26b7ea560c02 100644
--- a/drivers/staging/media/atomisp/pci/sh_css.c
+++ b/drivers/staging/media/atomisp/pci/sh_css.c
@@ -5819,36 +5819,37 @@ static int ia_css_pipe_create_cas_scaler_desc_single_output(
 		i *= max_scale_factor_per_stage;
 	}

-	descr->in_info = kmalloc(descr->num_stage *
-				 sizeof(struct ia_css_frame_info),
-				 GFP_KERNEL);
+	descr->in_info = kmalloc_objs(*descr->in_info,
+				      descr->num_stage,
+				      GFP_KERNEL);
 	if (!descr->in_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->internal_out_info = kmalloc(descr->num_stage *
-					   sizeof(struct ia_css_frame_info),
-					   GFP_KERNEL);
+	descr->internal_out_info = kmalloc_objs(*descr->internal_out_info,
+						descr->num_stage,
+						GFP_KERNEL);
 	if (!descr->internal_out_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->out_info = kmalloc(descr->num_stage *
-				  sizeof(struct ia_css_frame_info),
-				  GFP_KERNEL);
+	descr->out_info = kmalloc_objs(*descr->out_info,
+				       descr->num_stage,
+				       GFP_KERNEL);
 	if (!descr->out_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->vf_info = kmalloc(descr->num_stage *
-				 sizeof(struct ia_css_frame_info),
-				 GFP_KERNEL);
+	descr->vf_info = kmalloc_objs(*descr->vf_info,
+				      descr->num_stage,
+				      GFP_KERNEL);
 	if (!descr->vf_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->is_output_stage = kmalloc(descr->num_stage * sizeof(bool),
-					 GFP_KERNEL);
+	descr->is_output_stage = kmalloc_objs(*descr->is_output_stage,
+					      descr->num_stage,
+					      GFP_KERNEL);
 	if (!descr->is_output_stage) {
 		err = -ENOMEM;
 		goto ERR;
@@ -5968,35 +5969,37 @@ ia_css_pipe_create_cas_scaler_desc(struct ia_css_pipe *pipe,

 	descr->num_stage = num_stages;

-	descr->in_info = kmalloc_objs(struct ia_css_frame_info,
-				      descr->num_stage);
+	descr->in_info = kmalloc_objs(*descr->in_info,
+				      descr->num_stage,
+				      GFP_KERNEL);
 	if (!descr->in_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->internal_out_info = kmalloc(descr->num_stage *
-					   sizeof(struct ia_css_frame_info),
-					   GFP_KERNEL);
+	descr->internal_out_info = kmalloc_objs(*descr->internal_out_info,
+						descr->num_stage,
+						GFP_KERNEL);
 	if (!descr->internal_out_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->out_info = kmalloc(descr->num_stage *
-				  sizeof(struct ia_css_frame_info),
-				  GFP_KERNEL);
+	descr->out_info = kmalloc_objs(*descr->out_info,
+				       descr->num_stage,
+				       GFP_KERNEL);
 	if (!descr->out_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->vf_info = kmalloc(descr->num_stage *
-				 sizeof(struct ia_css_frame_info),
-				 GFP_KERNEL);
+	descr->vf_info = kmalloc_objs(*descr->vf_info,
+				      descr->num_stage,
+				      GFP_KERNEL);
 	if (!descr->vf_info) {
 		err = -ENOMEM;
 		goto ERR;
 	}
-	descr->is_output_stage = kmalloc(descr->num_stage * sizeof(bool),
-					 GFP_KERNEL);
+	descr->is_output_stage = kmalloc_objs(*descr->is_output_stage,
+					      descr->num_stage,
+					      GFP_KERNEL);
 	if (!descr->is_output_stage) {
 		err = -ENOMEM;
 		goto ERR;
--
2.53.0

Re: [PATCH] media: atomisp: use kmalloc_objs for array allocations

[Andy Shevchenko]

On Thu, Apr 16, 2026 at 4:46 PM Pedro Pontes <pontescpedro@gmail.com> wrote:
>
> Convert manual kmalloc() multiplications to the modern kmalloc_objs()
> interface to improve type safety and prevent potential integer
> overflows.

There is already a patch doing it in a slightly better way. Have you
followed the mailing list?
Please, better to help with this driver is to subscribe to the mailing
list and review already
submitted ones.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] media: atomisp: use kmalloc_objs for array allocations

[Pedro Pontes]

> There is already a patch doing it in a slightly better way. Have you
> followed the mailing list?

I searched lore and patchwork but couldn’t find the patch you’re referring to.
Could you please point me to it?

Are you referring to the patch I linked in my submission, or a different one?

> Please, better to help with this driver is to subscribe to the mailing
> list and review already
> submitted ones.

Understood, I’ll follow the list more closely and review related patches.
Thanks.

Re: [PATCH] media: atomisp: use kmalloc_objs for array allocations

[Sakari Ailus]

Hi Andy,

On Thu, Apr 16, 2026 at 09:26:47PM +0300, Andy Shevchenko wrote:
> On Thu, Apr 16, 2026 at 4:46 PM Pedro Pontes <pontescpedro@gmail.com> wrote:
> >
> > Convert manual kmalloc() multiplications to the modern kmalloc_objs()
> > interface to improve type safety and prevent potential integer
> > overflows.
> 
> There is already a patch doing it in a slightly better way. Have you
> followed the mailing list?

I must have missed it, too. :-\

> Please, better to help with this driver is to subscribe to the mailing
> list and review already
> submitted ones.

-- 
Regards,

Sakari Ailus

Re: [PATCH] media: atomisp: use kmalloc_objs for array allocations

[Andy Shevchenko]

On Tue, Apr 21, 2026 at 7:52 PM Sakari Ailus
<sakari.ailus@linux.intel.com> wrote:
> On Thu, Apr 16, 2026 at 09:26:47PM +0300, Andy Shevchenko wrote:
> > On Thu, Apr 16, 2026 at 4:46 PM Pedro Pontes <pontescpedro@gmail.com> wrote:
> > >
> > > Convert manual kmalloc() multiplications to the modern kmalloc_objs()
> > > interface to improve type safety and prevent potential integer
> > > overflows.
> >
> > There is already a patch doing it in a slightly better way. Have you
> > followed the mailing list?
>
> I must have missed it, too. :-\

https://lore.kernel.org/all/CAHp75VcoNkEQs7QQLDg8xZjdouNc8Yc6Vjg+Pdqb7LViqSKysg@mail.gmail.com/

> > Please, better to help with this driver is to subscribe to the mailing
> > list and review already
> > submitted ones.

-- 
With Best Regards,
Andy Shevchenko

Re: [PATCH] media: atomisp: use kmalloc_objs for array allocations

[Andy Shevchenko]

On Tue, Apr 21, 2026 at 9:07 PM Andy Shevchenko
<andy.shevchenko@gmail.com> wrote:
> On Tue, Apr 21, 2026 at 7:52 PM Sakari Ailus
> <sakari.ailus@linux.intel.com> wrote:
> > On Thu, Apr 16, 2026 at 09:26:47PM +0300, Andy Shevchenko wrote:
> > > On Thu, Apr 16, 2026 at 4:46 PM Pedro Pontes <pontescpedro@gmail.com> wrote:
> > > >
> > > > Convert manual kmalloc() multiplications to the modern kmalloc_objs()
> > > > interface to improve type safety and prevent potential integer
> > > > overflows.
> > >
> > > There is already a patch doing it in a slightly better way. Have you
> > > followed the mailing list?
> >
> > I must have missed it, too. :-\
>
> https://lore.kernel.org/all/CAHp75VcoNkEQs7QQLDg8xZjdouNc8Yc6Vjg+Pdqb7LViqSKysg@mail.gmail.com/

Hmm... Now re-reading that it seems they address semantically the same
issue, but for different APIs.

> > > Please, better to help with this driver is to subscribe to the mailing
> > > list and review already
> > > submitted ones.


-- 
With Best Regards,
Andy Shevchenko