Re: [PATCH] crypto: octeontx2: fix potential null pointer access

[Dan Carpenter <dan.carpenter@oracle.com>]

On Fri, May 27, 2022 at 01:27:56PM +0530, Shijith Thotton wrote:
> diff --git a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
> index 9cba2f714c7e..b91401929fc6 100644
> --- a/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
> +++ b/drivers/crypto/marvell/octeontx2/otx2_cptpf_ucode.c
> @@ -1605,7 +1605,11 @@ int otx2_cpt_dl_custom_egrp_create(struct otx2_cptpf_dev *cptpf,
>  		if (!strncasecmp(val, "se", 2) && strchr(val, ':')) {
>  			if (has_se || ucode_idx)
>  				goto err_print;
> -			tmp = strim(strsep(&val, ":"));
> +			tmp = strsep(&val, ":");
> +			if (tmp != NULL)
> +				tmp = strim(tmp);
> +			else
> +				goto err_print;

The check is not needed here, but if it were then the better way to
write this would be:

		tmp = strsep(&val, ":");
		if (!tmp)
			goto err_print;
		tmp = strim(tmp);

Always to error handling, not success handling.  checkpatch.pl --strict
will complain about the != NULL.

regards,
dan carpenter