Re: [dpdk-stable] [PATCH v9] ethdev: add sanity checks in control APIs

[Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>]

On 4/21/21 5:36 AM, Ferruh Yigit wrote:
> From: "Min Hu (Connor)" <humin29@huawei.com>
> 
> This patch adds more sanity checks in control path APIs.
> 
> Fixes: 214ed1acd125 ("ethdev: add iterator to match devargs input")
> Fixes: 3d98f921fbe9 ("ethdev: unify prefix for static functions and variables")
> Fixes: 0366137722a0 ("ethdev: check for invalid device name")
> Fixes: d948f596fee2 ("ethdev: fix port data mismatched in multiple process model")
> Fixes: 5b7ba31148a8 ("ethdev: add port ownership")
> Fixes: f8244c6399d9 ("ethdev: increase port id range")
> Cc: stable@dpdk.org
> 
> Signed-off-by: Min Hu (Connor) <humin29@huawei.com>
> Signed-off-by: Ferruh Yigit <ferruh.yigit@intel.com>
> Reviewed-by: Andrew Rybchenko <andrew.rybchenko@oktetlabs.ru>
> Acked-by: Kevin Traynor <ktraynor@redhat.com>

Few nits below.
Other than that I confirm my "Reviewed-by".

The patch is really long. It would be better to split it into
few:
 - relocate dev assignment
 - empty lines mangling (when it is unrelated to previous item)
 - ops check before usage (combined with related style checks)
 - error logs refinement

However, since the patch is already reviewed this way, may
be it is better to push as is after review notes processing.

> @@ -817,7 +859,12 @@ rte_eth_dev_get_port_by_name(const char *name, uint16_t *port_id)
>  	uint16_t pid;
>  
>  	if (name == NULL) {
> -		RTE_ETHDEV_LOG(ERR, "Null pointer is specified\n");
> +		RTE_ETHDEV_LOG(ERR, "Cannot get port ID from NULL name");
> +		return -EINVAL;
> +	}
> +
> +	if (port_id == NULL) {
> +		RTE_ETHDEV_LOG(ERR, "Cannot get port ID to NULL\n");

Since name is already checked above, I think it would be useful
to log 'name' here to provide context.

>  		return -EINVAL;
>  	}
>  

[snip]

> @@ -3256,6 +3370,20 @@ rte_eth_dev_fw_version_get(uint16_t port_id, char *fw_version, size_t fw_size)
>  	RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
>  	dev = &rte_eth_devices[port_id];
>  
> +	if (fw_version == NULL) {
> +		RTE_ETHDEV_LOG(ERR,
> +			"Cannot get ethdev port %u FW version to NULL\n",
> +			port_id);
> +		return -EINVAL;
> +	}
> +
> +	if (fw_size == 0) {
> +		RTE_ETHDEV_LOG(ERR,
> +			"Cannot get ethdev port %u FW version to buffer with zero size\n",
> +			port_id);
> +		return -EINVAL;
> +	}
> +

The only error condition is NULL fw_version with positive
fw_size. Othwerwise, it could be just a call to get required
size of buffer for FW version.

>  	RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->fw_version_get, -ENOTSUP);
>  	return eth_err(port_id, (*dev->dev_ops->fw_version_get)(dev,
>  							fw_version, fw_size));

[snip]

> @@ -3323,6 +3457,21 @@ rte_eth_dev_get_supported_ptypes(uint16_t port_id, uint32_t ptype_mask,
>  
>  	RTE_ETH_VALID_PORTID_OR_ERR_RET(port_id, -ENODEV);
>  	dev = &rte_eth_devices[port_id];
> +
> +	if (ptypes == NULL) {
> +		RTE_ETHDEV_LOG(ERR,
> +			"Cannot get ethdev port %u supported packet types to NULL\n",
> +			port_id);
> +		return -EINVAL;
> +	}
> +
> +	if (num == 0) {
> +		RTE_ETHDEV_LOG(ERR,
> +			"Cannot get ethdev port %u supported packet types to array with zero size\n",
> +			port_id);
> +		return -EINVAL;
> +	}
> +

The error condition is "ptypes == NULL && num > 0".
Otherwise, if num == 0 (regardless ptypes) it is just
a call to get size of required buffer. Same as below.

>  	RTE_FUNC_PTR_OR_ERR_RET(*dev->dev_ops->dev_supported_ptypes_get, 0);
>  	all_ptypes = (*dev->dev_ops->dev_supported_ptypes_get)(dev);
>  

[snip]