[PATCH] ASoC: topology: Fix memory corruption in soc_tplg_denum_create_values()

[Amadeusz Sławiński]

On 1/20/2021 10:59 AM, Dan Carpenter wrote:
> The allocation uses sizeof(u32) when it should use sizeof(unsigned long)
> so it leads to memory corruption later in the function when the data is
> initialized.
> 
> Fixes: 5aebe7c7f9c2 ("ASoC: topology: fix endianness issues")
> Signed-off-by: Dan Carpenter <dan.carpenter at oracle.com>
> ---
> This is from static analysis, not from testing.  Obviously we don't
> want memory corruption, so my patch is an improvement.  But I feel like
> a better approach might be to change the type of dvalues[] to u32.  I
> took the less risky approach because I'm not an expert and can't test
> it.  But if someone else can take a look at it, then I'll redo the
> patch.
> 
>   sound/soc/soc-topology.c | 2 +-
>   1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/sound/soc/soc-topology.c b/sound/soc/soc-topology.c
> index 950c45008e24..37a5d73e643b 100644
> --- a/sound/soc/soc-topology.c
> +++ b/sound/soc/soc-topology.c
> @@ -902,7 +902,7 @@ static int soc_tplg_denum_create_values(struct soc_tplg *tplg, struct soc_enum *
>   		return -EINVAL;
>   
>   	se->dobj.control.dvalues = devm_kcalloc(tplg->dev, le32_to_cpu(ec->items),
> -					   sizeof(u32),
> +					   sizeof(*se->dobj.control.dvalues),
>   					   GFP_KERNEL);
>   	if (!se->dobj.control.dvalues)
>   		return -ENOMEM;
> 

Looks good to me. And yes as we store already parsed value, dvalues 
could be changed to u32, but I would still change the sizeof as you did 
above.