Re: [dpdk-dev] [PATCH] Added Spinlock to l3fwd-vf example to prevent race conditioning

["Xie, Huawei" <huawei.xie@intel.com>]

> -----Original Message-----
> From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Thomas Monjalon
> Sent: Tuesday, November 11, 2014 3:57 PM
> To: Mrzyglod, DanielX T
> Cc: dev@dpdk.org
> Subject: Re: [dpdk-dev] [PATCH] Added Spinlock to l3fwd-vf example to prevent
> race conditioning
> 
> Hi Daniel,
> 
> This old patch is probably good but I'd like you explain it please.
> Reviewers are also welcome.
> 
> Thanks
> --
> Thomas
> 
> 2014-07-23 10:33, Thomas Monjalon:
> > Hi Daniel,
> >
> > Some explanations are missing here.
> >
> > > Signed-off-by: Daniel Mrzyglod <danielx.t.mrzyglod@intel.com>
> > >
> > > --- a/examples/l3fwd-vf/main.c
> > > +++ b/examples/l3fwd-vf/main.c
> > > @@ -54,6 +54,7 @@
> > >  #include <rte_per_lcore.h>
> > >  #include <rte_launch.h>
> > >  #include <rte_atomic.h>
> > > +#include <rte_spinlock.h>
> > >  #include <rte_cycles.h>
> > >  #include <rte_prefetch.h>
> > >  #include <rte_lcore.h>
> > > @@ -328,7 +329,7 @@ struct lcore_conf {
> > >  } __rte_cache_aligned;
> > >
> > >  static struct lcore_conf lcore_conf[RTE_MAX_LCORE];
> > > -
> > > +static rte_spinlock_t
> spinlock_conf[RTE_MAX_ETHPORTS]={RTE_SPINLOCK_INITIALIZER};
> > >  /* Send burst of packets on an output interface */
> > >  static inline int
> > >  send_burst(struct lcore_conf *qconf, uint16_t n, uint8_t port)
> > > @@ -340,7 +341,10 @@ send_burst(struct lcore_conf *qconf, uint16_t n,
> uint8_t port)
> > >  	queueid = qconf->tx_queue_id;
> > >  	m_table = (struct rte_mbuf **)qconf->tx_mbufs[port].m_table;
> > >
> > > +	rte_spinlock_lock(&spinlock_conf[port]) ;
> > >  	ret = rte_eth_tx_burst(port, queueid, m_table, n);
> > > +	rte_spinlock_unlock(&spinlock_conf[port]);

It might not be good choice for here, but how about we also provide spin_trylock as alternative API?

> > > +
> > >  	if (unlikely(ret < n)) {
> > >  		do {
> > >  			rte_pktmbuf_free(m_table[ret]);
> > >