Re: [PATCH v4] eventdev: add errno-style return values

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

On Tue, Mar 21, 2017 at 08:38:25PM +0000, Eads, Gage wrote:
> 
> 
> >  -----Original Message-----
> >  From: Jerin Jacob [mailto:jerin.jacob@caviumnetworks.com]
> >  Sent: Tuesday, March 21, 2017 6:06 AM
> >  To: Eads, Gage <gage.eads@intel.com>
> >  Cc: dev@dpdk.org; Richardson, Bruce <bruce.richardson@intel.com>;
> >  hemant.agrawal@nxp.com; Van Haaren, Harry <harry.van.haaren@intel.com>;
> >  nipun.gupta@nxp.com
> >  Subject: Re: [PATCH v4] eventdev: add errno-style return values
> >  
> >  On Fri, Mar 17, 2017 at 09:51:28AM -0500, Gage Eads wrote:
> >  > From: "Eads, Gage" <gage.eads@intel.com>
> >  >
> >  > This commit adds rte_errno return values to rte_event_enqueue_burst()
> >  > and rte_event_dequeue_burst().
> >  >
> >  > These return values allows user software to differentiate between an
> >  > invalid argument (such as an invalid queue_id or sched_type in an
> >  > enqueued
> >  > event) and backpressure from the event device.
> >  >
> >  > The port and device ID checks are placed in RTE_LIBRTE_EVENTDEV_DEBUG
> >  > header guards to avoid the performance hit in non-debug execution.
> >  >
> >  > Signed-off-by: Gage Eads <gage.eads@intel.com>
> >  > ---
> >  > Changes for v2:
> >  >   - Remove rte_errno initialization
> >  > Changes for v3:
> >  >   - Fix checkpatch and check-git-log.sh errors Changes for v4:
> >  >   - v3 was incorrectly based on v1, v4 is instead based on v2's
> >  > changes
> >  >
> >  >  lib/librte_eventdev/rte_eventdev.h | 40
> >  > +++++++++++++++++++++++++++++++++++---
> >  >  1 file changed, 37 insertions(+), 3 deletions(-)
> >  >
> >  > diff --git a/lib/librte_eventdev/rte_eventdev.h
> >  > b/lib/librte_eventdev/rte_eventdev.h
> >  > index 2b30a35..a45c81a 100644
> >  > --- a/lib/librte_eventdev/rte_eventdev.h
> >  > +++ b/lib/librte_eventdev/rte_eventdev.h
> >  > @@ -245,6 +245,7 @@ extern "C" {
> >  >
> >  >  #include <rte_common.h>
> >  >  #include <rte_memory.h>
> >  > +#include <rte_errno.h>
> >  >
> >  >  struct rte_mbuf; /* we just use mbuf pointers; no need to include
> >  > rte_mbuf.h */
> >  >
> >  > @@ -1118,9 +1119,14 @@ rte_event_schedule(uint8_t dev_id)
> >  >   *   The number of event objects actually enqueued on the event device. The
> >  >   *   return value can be less than the value of the *nb_events* parameter
> >  when
> >  >   *   the event devices queue is full or if invalid parameters are specified in a
> >  > - *   *rte_event*. If return value is less than *nb_events*, the remaining
> >  events
> >  > - *   at the end of ev[] are not consumed,and the caller has to take care of
> >  them
> >  > - *
> >  > + *   *rte_event*. If the return value is less than *nb_events*, the remaining
> >  > + *   events at the end of ev[] are not consumed and the caller has to take
> >  care
> >  > + *   of them, and rte_errno is set accordingly. Possible errno values include:
> >  > + *   -(-EINVAL) The port ID is invalid, device ID is invalid, an event's queue
> >  > + *              ID is invalid, or an event's sched type doesn't match the
> >  > + *              capabilities of the destination queue.
> >  > + *   -(-ENOSPC) The event port was backpressured and unable to enqueue
> >  > + *              one or more events.
> >  
> >  Some reason you haven't addressed the comments in v2.i.e add a note that -
> >  ENOSPC applicable only for *closed system*
> >  
> >  http://dpdk.org/ml/archives/dev/2017-March/060352.html
> >  
> 
> Will fix.
> 
> >  
> >  >   * @see rte_event_port_enqueue_depth()
> >  >   */
> >  >  static inline uint16_t
> >  > @@ -1129,6 +1135,20 @@ rte_event_enqueue_burst(uint8_t dev_id, uint8_t
> >  > port_id,  {
> >  >  	struct rte_eventdev *dev = &rte_eventdevs[dev_id];
> >  >
> >  > +#ifdef RTE_LIBRTE_EVENTDEV_DEBUG
> >  
> >  Some reason you haven't addressed the comments in v2.i.e Not fixed the build
> >  issue.
> >  http://dpdk.org/ml/archives/dev/2017-March/060352.html
> >  
> >  You can reproduce it by setting CONFIG_RTE_LIBRTE_EVENTDEV_DEBUG=y in
> >  common config.
> >  
> >  We have two approaches to fix it
> >  1/ Pollute lib/librte_eventdev/rte_eventdev.h header with implementation
> >  specific header files
> >  
> >  Or
> >  
> >  2/ Have neat pmd debug common function in rte_eventdev_pmd.h.Let all PMD
> >  driver call it by including rte_eventdev_pmd.h
> >  
> 
> I'd prefer to avoid the issue by dropping the debug message and RTE_EVENTDEV_DETACHED, like so:
> 
> #ifdef RTE_LIBRTE_EVENTDEV_DEBUG
>         if (dev_id >= RTE_EVENT_MAX_DEVS || !rte_eventdevs[dev_id].attached) {
>                 rte_errno = -EINVAL;
>                 return 0;
>         }
> 
>         if (port_id >= dev->data->nb_ports) {
>                 rte_errno = -EINVAL;
>                 return 0;
>         }
> #endif
> 
> What do you think?

Looks good to me.