From: Hemant Agrawal <hemant.agrawal@nxp.com>
To: "Hunt, David" <david.hunt@intel.com>,
"olivier.matz@6wind.com" <olivier.matz@6wind.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>,
"jerin.jacob@caviumnetworks.com" <jerin.jacob@caviumnetworks.com>
Subject: Re: [dpdk-dev] [PATCH v2 1/2] eal/mempool: introduce check for external mempool availability
Date: Fri, 16 Sep 2016 09:34:01 +0000 [thread overview]
Message-ID: <DB5PR04MB1605AC94C13CE93C0768687F89F30@DB5PR04MB1605.eurprd04.prod.outlook.com> (raw)
In-Reply-To: <f6f72a80-f6ec-ef10-c60b-f59b47bf1771@intel.com>
HI David,
> -----Original Message-----
> From: Hunt, David [mailto:david.hunt@intel.com]
> Sent: Friday, September 16, 2016 2:44 PM
> To: Hemant Agrawal <hemant.agrawal@nxp.com>; olivier.matz@6wind.com
> Cc: dev@dpdk.org; jerin.jacob@caviumnetworks.com
> Subject: Re: [PATCH v2 1/2] eal/mempool: introduce check for external
> mempool availability
>
> Hi Hemant,
>
> On 15/9/2016 6:13 PM, Hemant Agrawal wrote:
> > External offloaded mempool may not be available always. This check
> > enables run time verification of the presence of external mempool
> > before the mempool ops are installed.
> >
> > An application built with a specific external mempool may work fine on
> > host. But, may not work on VM, specificaly if non-hw specific
> > interfaces are used e.g. virtio-net.
> >
> > This is required to make sure that same application can work in all
> > environment for a given hw platform.
> >
> > The solution proposed here is that rte_mempool_set_ops_byname should
> > return specific error in case the current execution environment is not
> > supporting the given mempool. Thereby allowing it to fallback on
> > software mempool implementation e.g. "ring_mp_mc"
> >
> > This patch introduces new optional "pool_verify" as mempool ops
> > function to check if external mempool instance is available or not.
>
> I've a small suggestion around the naming of the new functions. It seems to me
> that this patch is more about mempool handler 'support' than 'verification'.
> Could I suggest a different name for this new function? I think maybe
> "supported" may be more appropriate than "pool_verify". The return code that's
> returned when a mempool handler is not available is -EOPNOTSUPP, which is
> what suggested the word "supported" to me. I think that:
>
> if (ops->supported)
> return ops->supported(mp);
>
> makes for slightly easier reading. The wrapper function would logically then be
> called rte_mempool_ops_supported().
[Hemant] Good suggestion. I will update it in next version.
>
>
> > Signed-off-by: Hemant Agrawal <hemant.agrawal@nxp.com>
> > ---
> > Changes in v2:
> > * fixed the pool_verify copy in rte_mempool_register_ops
> > ---
> > lib/librte_mempool/rte_mempool.h | 21 +++++++++++++++++++++
> > lib/librte_mempool/rte_mempool_ops.c | 19 +++++++++++++++++++
> > lib/librte_mempool/rte_mempool_ring.c | 4 ++++
> > lib/librte_mempool/rte_mempool_stack.c | 3 ++-
> > 4 files changed, 46 insertions(+), 1 deletion(-)
> >
> > diff --git a/lib/librte_mempool/rte_mempool.h
> b/lib/librte_mempool/rte_mempool.h
> > index 0243f9e..8599df1 100644
> > --- a/lib/librte_mempool/rte_mempool.h
> > +++ b/lib/librte_mempool/rte_mempool.h
> > @@ -387,6 +387,12 @@ typedef int (*rte_mempool_dequeue_t)(struct
> rte_mempool *mp,
> > */
> > typedef unsigned (*rte_mempool_get_count)(const struct rte_mempool
> *mp);
> >
> > +/**
> > + * Return if the given external mempool is available for this instance.
> > + * it is optional to implement for mempools
> > + */
> > +typedef int (*rte_mempool_pool_verify)(const struct rte_mempool *mp);
> > +
> > /** Structure defining mempool operations structure */
> > struct rte_mempool_ops {
> > char name[RTE_MEMPOOL_OPS_NAMESIZE]; /**< Name of mempool
> ops struct. */
> > @@ -395,6 +401,8 @@ struct rte_mempool_ops {
> > rte_mempool_enqueue_t enqueue; /**< Enqueue an object. */
> > rte_mempool_dequeue_t dequeue; /**< Dequeue an object. */
> > rte_mempool_get_count get_count; /**< Get qty of available objs. */
> > + rte_mempool_pool_verify pool_verify;
> > + /**< Verify if external mempool is available for usages*/
> > } __rte_cache_aligned;
> >
> > #define RTE_MEMPOOL_MAX_OPS_IDX 16 /**< Max registered ops structs
> */
> > @@ -516,6 +524,18 @@ void
> > rte_mempool_ops_free(struct rte_mempool *mp);
> >
> > /**
> > + * @internal wrapper to verify the external mempool availability
> > + *
> > + * @param mp
> > + * Pointer to the memory pool.
> > + * @return
> > + * 0: Success; external mempool instance is available
> > + * - <0: Error; external mempool instance is not available
> > + */
> > +int
> > +rte_mempool_ops_pool_verify(const struct rte_mempool *mp);
> > +
> > +/**
> > * Set the ops of a mempool.
> > *
> > * This can only be done on a mempool that is not populated, i.e. just after
> > @@ -531,6 +551,7 @@ rte_mempool_ops_free(struct rte_mempool *mp);
> > * - 0: Success; the mempool is now using the requested ops functions.
> > * - -EINVAL - Invalid ops struct name provided.
> > * - -EEXIST - mempool already has an ops struct assigned.
> > + * - -EOPNOTSUPP - mempool instance not available.
> > */
> > int
> > rte_mempool_set_ops_byname(struct rte_mempool *mp, const char *name,
> > diff --git a/lib/librte_mempool/rte_mempool_ops.c
> b/lib/librte_mempool/rte_mempool_ops.c
> > index 5f24de2..c2e765f 100644
> > --- a/lib/librte_mempool/rte_mempool_ops.c
> > +++ b/lib/librte_mempool/rte_mempool_ops.c
> > @@ -85,6 +85,7 @@ rte_mempool_register_ops(const struct
> rte_mempool_ops *h)
> > ops->enqueue = h->enqueue;
> > ops->dequeue = h->dequeue;
> > ops->get_count = h->get_count;
> > + ops->pool_verify = h->pool_verify;
> >
> > rte_spinlock_unlock(&rte_mempool_ops_table.sl);
> >
> > @@ -123,6 +124,18 @@ rte_mempool_ops_get_count(const struct
> rte_mempool *mp)
> > return ops->get_count(mp);
> > }
> >
> > +/* wrapper to check if given external mempool is available for this
> instance.*/
> > +int
> > +rte_mempool_ops_pool_verify(const struct rte_mempool *mp)
> > +{
> > + struct rte_mempool_ops *ops;
> > +
> > + ops = rte_mempool_get_ops(mp->ops_index);
> > + if (ops->pool_verify)
> > + return ops->pool_verify(mp);
> > + return 0;
> > +}
> > +
> > /* sets mempool ops previously registered by rte_mempool_register_ops. */
> > int
> > rte_mempool_set_ops_byname(struct rte_mempool *mp, const char *name,
> > @@ -146,6 +159,12 @@ rte_mempool_set_ops_byname(struct rte_mempool
> *mp, const char *name,
> > if (ops == NULL)
> > return -EINVAL;
> >
> > + /*verify if the given mempool is available for this instance */
> > + if (ops->pool_verify) {
> > + if (ops->pool_verify(mp))
> > + return -EOPNOTSUPP;
> > + }
> > +
> > mp->ops_index = i;
> > mp->pool_config = pool_config;
> > return 0;
> > diff --git a/lib/librte_mempool/rte_mempool_ring.c
> b/lib/librte_mempool/rte_mempool_ring.c
> > index b9aa64d..d86d5e0 100644
> > --- a/lib/librte_mempool/rte_mempool_ring.c
> > +++ b/lib/librte_mempool/rte_mempool_ring.c
> > @@ -126,6 +126,7 @@ static const struct rte_mempool_ops ops_mp_mc = {
> > .enqueue = common_ring_mp_enqueue,
> > .dequeue = common_ring_mc_dequeue,
> > .get_count = common_ring_get_count,
> > + .pool_verify = NULL,
> > };
> >
> > static const struct rte_mempool_ops ops_sp_sc = {
> > @@ -135,6 +136,7 @@ static const struct rte_mempool_ops ops_sp_sc = {
> > .enqueue = common_ring_sp_enqueue,
> > .dequeue = common_ring_sc_dequeue,
> > .get_count = common_ring_get_count,
> > + .pool_verify = NULL,
> > };
> >
> > static const struct rte_mempool_ops ops_mp_sc = {
> > @@ -144,6 +146,7 @@ static const struct rte_mempool_ops ops_mp_sc = {
> > .enqueue = common_ring_mp_enqueue,
> > .dequeue = common_ring_sc_dequeue,
> > .get_count = common_ring_get_count,
> > + .pool_verify = NULL,
> > };
> >
> > static const struct rte_mempool_ops ops_sp_mc = {
> > @@ -153,6 +156,7 @@ static const struct rte_mempool_ops ops_sp_mc = {
> > .enqueue = common_ring_sp_enqueue,
> > .dequeue = common_ring_mc_dequeue,
> > .get_count = common_ring_get_count,
> > + .pool_verify = NULL,
> > };
> >
> > MEMPOOL_REGISTER_OPS(ops_mp_mc);
> > diff --git a/lib/librte_mempool/rte_mempool_stack.c
> b/lib/librte_mempool/rte_mempool_stack.c
> > index 5fd8af2..0198b70 100644
> > --- a/lib/librte_mempool/rte_mempool_stack.c
> > +++ b/lib/librte_mempool/rte_mempool_stack.c
> > @@ -141,7 +141,8 @@ static struct rte_mempool_ops ops_stack = {
> > .free = stack_free,
> > .enqueue = stack_enqueue,
> > .dequeue = stack_dequeue,
> > - .get_count = stack_get_count
> > + .get_count = stack_get_count,
> > + .pool_verify = NULL,
> > };
> >
> > MEMPOOL_REGISTER_OPS(ops_stack);
>
> Regards,
> Dave.
next prev parent reply other threads:[~2016-09-16 9:34 UTC|newest]
Thread overview: 17+ messages / expand[flat|nested] mbox.gz Atom feed top
2016-09-08 14:50 [dpdk-dev] [PATCH " Hemant Agrawal
2016-09-08 14:50 ` [dpdk-dev] [PATCH 2/2] mempool:pktmbuf pool default fallback for mempool ops error Hemant Agrawal
2016-09-15 17:13 ` [dpdk-dev] [PATCH v2 1/2] eal/mempool: introduce check for external mempool availability Hemant Agrawal
2016-09-15 17:13 ` [dpdk-dev] [PATCH v2 2/2] mempool:pktmbuf pool default fallback for mempool ops error Hemant Agrawal
2016-09-16 8:29 ` Hunt, David
2016-09-16 9:34 ` Hemant Agrawal
2016-09-16 9:13 ` [dpdk-dev] [PATCH v2 1/2] eal/mempool: introduce check for external mempool availability Hunt, David
2016-09-16 9:34 ` Hemant Agrawal [this message]
2016-09-16 16:46 ` [dpdk-dev] [PATCH v3 1/2] eal/mempool: check for external mempool support Hemant Agrawal
2016-09-16 16:46 ` [dpdk-dev] [PATCH v3 2/2] mempool: pktmbuf pool default fallback for mempool ops error Hemant Agrawal
2016-09-19 13:57 ` Olivier Matz
2016-09-22 13:12 ` Hemant Agrawal
2016-10-13 13:15 ` Hemant Agrawal
2016-10-14 12:10 ` Olivier Matz
2016-11-07 12:30 ` Hemant Agrawal
2016-11-22 9:24 ` Olivier Matz
2016-12-08 8:57 ` Hemant Agrawal
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=DB5PR04MB1605AC94C13CE93C0768687F89F30@DB5PR04MB1605.eurprd04.prod.outlook.com \
--to=hemant.agrawal@nxp.com \
--cc=david.hunt@intel.com \
--cc=dev@dpdk.org \
--cc=jerin.jacob@caviumnetworks.com \
--cc=olivier.matz@6wind.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).