Re: [PATCH] HID: bpf: One function call less in call_hid_bpf_rdesc_fixup() after error detection

[Hou Tao]

Hi,

On 12/27/2023 2:24 AM, Markus Elfring wrote:
> From: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
> Date: Tue, 26 Dec 2023 19:13:25 +0100
>
> The kfree() function was called in one case by the
> call_hid_bpf_rdesc_fixup() function during error handling
> even if the passed data structure member contained a null pointer.
> This issue was detected by using the Coccinelle software.

It is totally OK to free a null pointer through kfree() and the ENOMEM
case is an unlikely case, so I don't think the patch is necessary.
>
> Thus adjust jump targets.
>
> Signed-off-by: Markus Elfring <elfring@xxxxxxxxxxxxxxxxxxxxx>
> ---
>  drivers/hid/bpf/hid_bpf_dispatch.c | 9 +++++----
>  1 file changed, 5 insertions(+), 4 deletions(-)
>
> diff --git a/drivers/hid/bpf/hid_bpf_dispatch.c b/drivers/hid/bpf/hid_bpf_dispatch.c
> index d9ef45fcaeab..c84fe55be5ed 100644
> --- a/drivers/hid/bpf/hid_bpf_dispatch.c
> +++ b/drivers/hid/bpf/hid_bpf_dispatch.c
> @@ -118,17 +118,17 @@ u8 *call_hid_bpf_rdesc_fixup(struct hid_device *hdev, u8 *rdesc, unsigned int *s
>
>  	ctx_kern.data = kzalloc(ctx_kern.ctx.allocated_size, GFP_KERNEL);
>  	if (!ctx_kern.data)
> -		goto ignore_bpf;
> +		goto dup_mem;
>
>  	memcpy(ctx_kern.data, rdesc, min_t(unsigned int, *size, HID_MAX_DESCRIPTOR_SIZE));
>
>  	ret = hid_bpf_prog_run(hdev, HID_BPF_PROG_TYPE_RDESC_FIXUP, &ctx_kern);
>  	if (ret < 0)
> -		goto ignore_bpf;
> +		goto free_data;
>
>  	if (ret) {
>  		if (ret > ctx_kern.ctx.allocated_size)
> -			goto ignore_bpf;
> +			goto free_data;
>
>  		*size = ret;
>  	}
> @@ -137,8 +137,9 @@ u8 *call_hid_bpf_rdesc_fixup(struct hid_device *hdev, u8 *rdesc, unsigned int *s
>
>  	return rdesc;
>
> - ignore_bpf:
> +free_data:
>  	kfree(ctx_kern.data);
> +dup_mem:
>  	return kmemdup(rdesc, *size, GFP_KERNEL);
>  }
>  EXPORT_SYMBOL_GPL(call_hid_bpf_rdesc_fixup);
> --
> 2.43.0
>
>
> .