[PATCH v2 18/18] media: venus: Refactor hfi_buffer_alloc_mode_supported

[Ricardo Ribalda]

Hi Bryan

On Fri, 10 May 2024 at 02:09, Bryan O'Donoghue
<bryan.odonoghue at linaro.org> wrote:
>
> On 07/05/2024 17:27, Ricardo Ribalda wrote:
> > Replace the old style single element array at the end of the struct with
> > a flex array.
> >
> > The code does not allocate this structure, so the size change should not
> > be a problem.
> >
> > This fixes the following cocci warning:
> > drivers/media/platform/qcom/venus/hfi_helper.h:1233:5-9: WARNING use flexible-array member instead (https://www.kernel.org/doc/html/latest/process/deprecated.html#zero-length-and-one-element-arrays)
> >
> > Signed-off-by: Ricardo Ribalda <ribalda at chromium.org>
> > ---
> >   drivers/media/platform/qcom/venus/hfi_helper.h | 2 +-
> >   1 file changed, 1 insertion(+), 1 deletion(-)
> >
> > diff --git a/drivers/media/platform/qcom/venus/hfi_helper.h b/drivers/media/platform/qcom/venus/hfi_helper.h
> > index dee439ea4d2e..9545c964a428 100644
> > --- a/drivers/media/platform/qcom/venus/hfi_helper.h
> > +++ b/drivers/media/platform/qcom/venus/hfi_helper.h
> > @@ -1230,7 +1230,7 @@ struct hfi_interlace_format_supported {
> >   struct hfi_buffer_alloc_mode_supported {
> >       u32 buffer_type;
> >       u32 num_entries;
> > -     u32 data[1];
> > +     u32 data[];
> >   };
> >
> >   struct hfi_metadata_pass_through {
> >
>
> You have some fairly inconsistent fixes for this class.
>
> Please don't change the sizeof() any structures in your series, because
> the structure is unallocated changing the size is potentially insidious IMO.

If the array is a flex array we should convert it to a real flex
array. Abusing one element arrays is deprecated.

If the driver only accesses the first element of the array, I have
modified the code from data[1] to data, because, as you say, modifying
the struct size can have dangerous side effects.

But if the driver accesses more fields, then I have reviewed that
there are no allocations, or any calculations based on the structure
side, and convert them to a proper flex array.


Regards!
>
> data[1] -> data is perfectly fine in this case.

If you take a look at parse_alloc_mode() you will see that mode->data
is indeed an array, and it is used by the pointer type.

>
> ---
> bod



-- 
Ricardo Ribalda