Re: [dpdk-dev, v1, 1/5] eventdev: add caps API and PMD callbacks for crypto adapter

[Jerin Jacob <jerin.jacob@caviumnetworks.com>]

-----Original Message-----
> Date: Wed, 4 Apr 2018 12:26:18 +0530
> From: Abhinandan Gujjar <abhinandan.gujjar@intel.com>
> To: jerin.jacob@caviumnetworks.com, hemant.agrawal@nxp.com,
>  akhil.goyal@nxp.com, dev@dpdk.org
> CC: pablo.de.lara.guarch@intel.com, declan.doherty@intel.com,
>  narender.vangati@intel.com, abhinandan.gujjar@intel.com,
>  nikhil.rao@intel.com
> Subject: [dpdk-dev,v1,1/5] eventdev: add caps API and PMD callbacks for
>  crypto adapter
> X-Mailer: git-send-email 1.9.1
> 
> Signed-off-by: Abhinandan Gujjar <abhinandan.gujjar@intel.com>
> ---
>  lib/librte_eventdev/rte_eventdev.c     |  25 +++++
>  lib/librte_eventdev/rte_eventdev.h     |  35 +++++++
>  lib/librte_eventdev/rte_eventdev_pmd.h | 176 +++++++++++++++++++++++++++++++++
>  3 files changed, 236 insertions(+)
> 
> diff --git a/lib/librte_eventdev/rte_eventdev.c b/lib/librte_eventdev/rte_eventdev.c
> index 2de8d9a..3d24e8f 100644
> --- a/lib/librte_eventdev/rte_eventdev.c
> +++ b/lib/librte_eventdev/rte_eventdev.c
> @@ -29,6 +29,8 @@
>  #include <rte_malloc.h>
>  #include <rte_errno.h>
>  #include <rte_ethdev.h>
> +#include <rte_cryptodev.h>
> +#include <rte_cryptodev_pmd.h>
>  
>  #include "rte_eventdev.h"
>  #include "rte_eventdev_pmd.h"
> @@ -123,6 +125,29 @@
>  				: 0;
>  }
>  
> +int __rte_experimental
> +rte_event_crypto_adapter_caps_get(uint8_t dev_id, uint8_t cdev_id,
> +				  uint32_t *caps)
> +{
> +	struct rte_eventdev *dev;
> +	struct rte_cryptodev *cdev;
> +
> +	RTE_EVENTDEV_VALID_DEVID_OR_ERR_RET(dev_id, -EINVAL);
> +	if (!rte_cryptodev_pmd_is_valid_dev(cdev_id))
> +		return -EINVAL;
> +
> +	dev = &rte_eventdevs[dev_id];
> +	cdev = rte_cryptodev_pmd_get_dev(cdev_id);
> +
> +	if (caps == NULL)
> +		return -EINVAL;
> +	*caps = 0;
> +
> +	return dev->dev_ops->crypto_adapter_caps_get ?
> +		(*dev->dev_ops->crypto_adapter_caps_get)
> +		(dev, cdev, caps) : 0;
> +}
> +
>  static inline int
>  rte_event_dev_queue_config(struct rte_eventdev *dev, uint8_t nb_queues)
>  {
> diff --git a/lib/librte_eventdev/rte_eventdev.h b/lib/librte_eventdev/rte_eventdev.h
> index a20077c..49a71d1 100644
> --- a/lib/librte_eventdev/rte_eventdev.h
> +++ b/lib/librte_eventdev/rte_eventdev.h
> @@ -35,6 +35,8 @@
>  #ifndef _RTE_EVENTDEV_H_
>  #define _RTE_EVENTDEV_H_
>  
> +#include <rte_compat.h>
> +
>  /**
>   * @file
>   *
> @@ -1142,6 +1144,39 @@ struct rte_event {
>  rte_event_eth_rx_adapter_caps_get(uint8_t dev_id, uint8_t eth_port_id,
>  				uint32_t *caps);
>  
> +
> +/* Crypto adapter capability bitmap flag */
> +#define RTE_EVENT_CRYPTO_ADAPTER_CAP_INTERNAL_PORT	0x1
> +/**< Flag indicates HW is capable of generating events.
> + * Cryptodev can send packets to the event device using an internal event port.
> + */

Some top level comments,

1) Since we are not planning to abstract RTE_EVENT_CRYPTO_ADAPTER_ENQ_DEQ
mode inside the adapter, I think, it make sense to introduce new
capability. ie. In some of the HW implementation, We could maintain the
packet order even if source queue/scheded_type is ORDERED. ie. if the capability
is set then application can use RTE_EVENT_CRYPTO_ADAPTER_DEQ_ONLY to maintain
the ingress order even if it is been called from multiple CPU in
ORDERED/ATOMIC context instead of using RTE_EVENT_CRYPTO_ADAPTER_ENQ_DEQ
mode.

2) Please split the 2/5 to patch specification and implementation

3) I think, we need to give a example code snippet in the programmer
guide on how to use RTE_EVENT_CRYPTO_ADAPTER_ENQ_DEQ mode in conjunction
with rte_event_crypto_adapter_event_port_get().

Let me know your views?