DPDK patches and discussions
 help / color / mirror / Atom feed
From: Vadim Suraev <vadim.suraev@gmail.com>
To: "Ananyev, Konstantin" <konstantin.ananyev@intel.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] [PATCH v2] rte_mbuf: mbuf bulk alloc/free functions added + unittest
Date: Wed, 18 Mar 2015 07:19:15 +0200
Message-ID: <CAJ0CJ8nR2DFe32ULB_J+c7aY_mBXxJnubkHYwX_=rwFRnE7NqQ@mail.gmail.com> (raw)
In-Reply-To: <2601191342CEEE43887BDE71AB977258213F6F10@irsmsx105.ger.corp.intel.com>

Hi, Konstantin,

>Shouldn't the line above be inside if (head != NULL) {...} block?

This is removed as Olivier commented before:

>> +{
> +     if (likely(head != NULL)) {

>I think we should remove this test. The other mbuf functions do not
>check this.

Regards,
 Vadim.

On Wed, Mar 18, 2015 at 1:46 AM, Ananyev, Konstantin <
konstantin.ananyev@intel.com> wrote:

> Hi Vadim,
>
> > -----Original Message-----
> > From: vadim.suraev@gmail.com [mailto:vadim.suraev@gmail.com]
> > Sent: Tuesday, March 17, 2015 9:36 PM
> > To: dev@dpdk.org
> > Cc: olivier.matz@6wind.com; stephen@networkplumber.org; Ananyev,
> Konstantin; vadim.suraev@gmail.com
> > Subject: [PATCH v2] rte_mbuf: mbuf bulk alloc/free functions added +
> unittest
> >
> > From: "vadim.suraev@gmail.com" <vadim.suraev@gmail.com>
> >
> > This patch adds mbuf bulk allocation/freeing functions and unittest
> >
> > Signed-off-by: Vadim Suraev
> > <vadim.suraev@gmail.com>
> > ---
> > New in v2:
> >     - function rte_pktmbuf_alloc_bulk added
> >     - function rte_pktmbuf_bulk_free added
> >     - function rte_pktmbuf_free_chain added
> >     - applied reviewers' comments
> >
> >  app/test/test_mbuf.c       |   94
> +++++++++++++++++++++++++++++++++++++++++++-
> >  lib/librte_mbuf/rte_mbuf.h |   89
> +++++++++++++++++++++++++++++++++++++++++
> >  2 files changed, 182 insertions(+), 1 deletion(-)
> >
> > diff --git a/app/test/test_mbuf.c b/app/test/test_mbuf.c
> > index 1ff66cb..b20c6a4 100644
> > --- a/app/test/test_mbuf.c
> > +++ b/app/test/test_mbuf.c
> > @@ -77,6 +77,7 @@
> >  #define REFCNT_RING_SIZE        (REFCNT_MBUF_NUM * REFCNT_MAX_REF)
> >
> >  #define MAKE_STRING(x)          # x
> > +#define MBUF_POOL_LOCAL_CACHE_SIZE 32
> >
> >  static struct rte_mempool *pktmbuf_pool = NULL;
> >
> > @@ -405,6 +406,84 @@ test_pktmbuf_pool(void)
> >       return ret;
> >  }
> >
> > +/* test pktmbuf bulk allocation and freeing
> > +*/
> > +static int
> > +test_pktmbuf_pool_bulk(void)
> > +{
> > +     unsigned i;
> > +     /* size of mempool - size of local cache, otherwise may fail */
> > +     unsigned mbufs_to_allocate = NB_MBUF - MBUF_POOL_LOCAL_CACHE_SIZE;
> > +     struct rte_mbuf *m[mbufs_to_allocate];
> > +     int ret = 0;
> > +     unsigned mbuf_count_before_allocation =
> rte_mempool_count(pktmbuf_pool);
> > +
> > +     for (i = 0; i < mbufs_to_allocate; i++)
> > +             m[i] = NULL;
> > +     /* alloc NB_MBUF-MBUF_POOL_LOCAL_CACHE_SIZE mbufs */
> > +     ret = rte_pktmbuf_alloc_bulk(pktmbuf_pool, m, mbufs_to_allocate);
> > +     if (ret) {
> > +             printf("cannot allocate %d mbufs bulk mempool_cnt=%d
> ret=%d\n",
> > +                     mbufs_to_allocate,
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                     ret);
> > +             return -1;
> > +     }
> > +     if ((rte_mempool_count(pktmbuf_pool) + mbufs_to_allocate) !=
> > +         mbuf_count_before_allocation) {
> > +             printf("mempool count %d + allocated %d != initial %d\n",
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                     mbufs_to_allocate,
> > +                     mbuf_count_before_allocation);
> > +             return -1;
> > +     }
> > +     /* free them */
> > +     rte_pktmbuf_bulk_free(m, mbufs_to_allocate);
> > +
> > +     if (rte_mempool_count(pktmbuf_pool)  !=
> mbuf_count_before_allocation) {
> > +             printf("mempool count %d != initial %d\n",
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                     mbuf_count_before_allocation);
> > +             return -1;
> > +     }
> > +     for (i = 0; i < mbufs_to_allocate; i++)
> > +             m[i] = NULL;
> > +
> > +     /* alloc NB_MBUF-MBUF_POOL_LOCAL_CACHE_SIZE mbufs */
> > +     ret = rte_pktmbuf_alloc_bulk(pktmbuf_pool, m, mbufs_to_allocate);
> > +     if (ret) {
> > +             printf("cannot allocate %d mbufs bulk mempool_cnt=%d
> ret=%d\n",
> > +                     mbufs_to_allocate,
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                     ret);
> > +             return -1;
> > +     }
> > +     if ((rte_mempool_count(pktmbuf_pool) + mbufs_to_allocate) !=
> > +         mbuf_count_before_allocation) {
> > +             printf("mempool count %d + allocated %d != initial %d\n",
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                                       mbufs_to_allocate,
> > +                                       mbuf_count_before_allocation);
> > +             return -1;
> > +     }
> > +
> > +     /* chain it */
> > +     for (i = 0; i < mbufs_to_allocate - 1; i++) {
> > +             m[i]->next = m[i + 1];
> > +             m[0]->nb_segs++;
> > +     }
> > +     /* free them */
> > +     rte_pktmbuf_free_chain(m[0]);
> > +
> > +     if (rte_mempool_count(pktmbuf_pool)  !=
> mbuf_count_before_allocation) {
> > +             printf("mempool count %d != initial %d\n",
> > +                     rte_mempool_count(pktmbuf_pool),
> > +                                       mbuf_count_before_allocation);
> > +             return -1;
> > +     }
> > +     return ret;
> > +}
> > +
> >  /*
> >   * test that the pointer to the data on a packet mbuf is set properly
> >   */
> > @@ -766,7 +845,8 @@ test_mbuf(void)
> >       if (pktmbuf_pool == NULL) {
> >               pktmbuf_pool =
> >                       rte_mempool_create("test_pktmbuf_pool", NB_MBUF,
> > -                                        MBUF_SIZE, 32,
> > +                                        MBUF_SIZE,
> > +                                        MBUF_POOL_LOCAL_CACHE_SIZE,
> >                                          sizeof(struct
> rte_pktmbuf_pool_private),
> >                                          rte_pktmbuf_pool_init, NULL,
> >                                          rte_pktmbuf_init, NULL,
> > @@ -790,6 +870,18 @@ test_mbuf(void)
> >               return -1;
> >       }
> >
> > +     /* test bulk allocation and freeing */
> > +     if (test_pktmbuf_pool_bulk() < 0) {
> > +             printf("test_pktmbuf_pool_bulk() failed\n");
> > +             return -1;
> > +     }
> > +
> > +     /* once again to ensure all mbufs were freed */
> > +     if (test_pktmbuf_pool_bulk() < 0) {
> > +             printf("test_pktmbuf_pool_bulk() failed\n");
> > +             return -1;
> > +     }
> > +
> >       /* test that the pointer to the data on a packet mbuf is set
> properly */
> >       if (test_pktmbuf_pool_ptr() < 0) {
> >               printf("test_pktmbuf_pool_ptr() failed\n");
> > diff --git a/lib/librte_mbuf/rte_mbuf.h b/lib/librte_mbuf/rte_mbuf.h
> > index 17ba791..995237d 100644
> > --- a/lib/librte_mbuf/rte_mbuf.h
> > +++ b/lib/librte_mbuf/rte_mbuf.h
> > @@ -825,6 +825,95 @@ static inline void rte_pktmbuf_free(struct rte_mbuf
> *m)
> >  }
> >
> >  /**
> > + * Allocate a bulk of mbufs, initiate refcnt and resets
> > + *
> > + * @param pool
> > + *    memory pool to allocate from
> > + * @param mbufs
> > + *    Array of pointers to mbuf
> > + * @param count
> > + *    Array size
> > + */
> > +static inline int rte_pktmbuf_alloc_bulk(struct rte_mempool *pool,
> > +                                      struct rte_mbuf **mbufs,
> > +                                      unsigned count)
> > +{
> > +     unsigned idx;
> > +     int rc = 0;
> > +
> > +     rc = rte_mempool_get_bulk(pool, (void **)mbufs, count);
> > +     if (unlikely(rc))
> > +             return rc;
> > +
> > +     for (idx = 0; idx < count; idx++) {
> > +             RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(mbufs[idx]) == 0);
> > +             rte_mbuf_refcnt_set(mbufs[idx], 1);
> > +             rte_pktmbuf_reset(mbufs[idx]);
> > +     }
> > +     return rc;
> > +}
> > +
> > +/**
> > + * Free a bulk of mbufs into its original mempool.
> > + * This function assumes refcnt equals 1
> > + * as well as the freed mbufs are direct
>
> I think your forgot to mention in comments one more requirement for that
> function:
> all mbufs have to be from the same mempool.
>
> > + *
> > + * @param mbufs
> > + *    Array of pointers to mbuf
> > + * @param count
> > + *    Array size
> > + */
> > +static inline void rte_pktmbuf_bulk_free(struct rte_mbuf **mbufs,
> > +                                      unsigned count)
> > +{
> > +     unsigned idx;
> > +
> > +     RTE_MBUF_ASSERT(count > 0);
> > +
> > +     for (idx = 0; idx < count; idx++) {
> > +             rte_mbuf_refcnt_update(mbufs[idx], -1);
>
> You can do just:
> rte_mbuf_refcnt_set(m, 0);
> here and move your assert above it.
> Something like:
> RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(mbufs[idx]) == 1);
> rte_mbuf_refcnt_set(m, 0);
>
> Also probably would be a good thing to add one more assert here,
> something like:
> RTE_MBUF_ASSERT(mbufs[idx]->pool == mufs[0]->pool);
>
> > +             RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(mbufs[idx]) == 0);
> > +     }
> > +     rte_mempool_put_bulk(mbufs[0]->pool, (void **)mbufs, count);
> > +}
> > +
> > +/**
> > + * Free chained (scattered) mbufs into its original mempool(s).
> > + *
> > + * @param head
> > + *    The head of mbufs to be freed chain. Must not be NULL
> > + */
> > +static inline void rte_pktmbuf_free_chain(struct rte_mbuf *head)
> > +{
> > +     struct rte_mbuf *mbufs[head->nb_segs];
> > +     unsigned mbufs_count = 0;
> > +     struct rte_mbuf *next;
> > +
> > +     while (head) {
> > +             next = head->next;
> > +             head->next = NULL;
>
> Shouldn't the line above be inside if (head != NULL) {...} block?
>
> > +             head = __rte_pktmbuf_prefree_seg(head);
> > +             if (likely(head != NULL)) {
> > +                     RTE_MBUF_ASSERT(rte_mbuf_refcnt_read(head) == 0);
>
> I don't think there is any use of the assert above.
> If prefree_seg returns non-NULL value, it sets refcnt to 0 for that mbuf.
>
> > +                     if (likely((!mbufs_count) ||
> > +                                (head->pool == mbufs[0]->pool)))
> > +                             mbufs[mbufs_count++] = head;
> > +                     else {
> > +                             rte_mempool_put_bulk(mbufs[0]->pool,
> > +                                                  (void **)mbufs,
> > +                                                  mbufs_count);
> > +                             mbufs_count = 0;
> > +                     }
> > +             }
> > +             head = next;
> > +     }
> > +     if (mbufs_count > 0)
> > +             rte_mempool_put_bulk(mbufs[0]->pool,
> > +                                  (void **)mbufs,
> > +                                  mbufs_count);
> > +}
> > +
> > +/**
> >   * Creates a "clone" of the given packet mbuf.
> >   *
> >   * Walks through all segments of the given packet mbuf, and for each of
> them:
> > --
> > 1.7.9.5
>
>

  reply	other threads:[~2015-03-18  5:19 UTC|newest]

Thread overview: 22+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2015-03-17 21:36 vadim.suraev
2015-03-17 23:46 ` Ananyev, Konstantin
2015-03-18  5:19   ` Vadim Suraev [this message]
     [not found]     ` <2601191342CEEE43887BDE71AB977258213F7053@irsmsx105.ger.corp.intel.com>
2015-03-18  9:56       ` Ananyev, Konstantin
2015-03-18 10:41         ` Vadim Suraev
     [not found]           ` <2601191342CEEE43887BDE71AB977258213F7136@irsmsx105.ger.corp.intel.com>
2015-03-18 15:13             ` Ananyev, Konstantin
2015-03-19  8:13               ` Olivier MATZ
2015-03-19 10:47                 ` Ananyev, Konstantin
2015-03-19 10:54                   ` Olivier MATZ
2015-03-18 20:21 vadim.suraev
2015-03-18 20:58 ` Neil Horman
2015-03-19  8:41   ` Olivier MATZ
2015-03-19 10:06     ` Ananyev, Konstantin
2015-03-19 13:16     ` Neil Horman
2015-03-23 16:44       ` Olivier MATZ
2015-03-23 17:31         ` Vadim Suraev
2015-03-23 23:48           ` Ananyev, Konstantin
2015-03-24  7:53             ` Vadim Suraev
     [not found]               ` <2601191342CEEE43887BDE71AB977258214071C0@irsmsx105.ger.corp.intel.com>
2015-03-24 11:00                 ` Ananyev, Konstantin
2015-03-23 18:45         ` Neil Horman
2015-03-30 19:04   ` Vadim Suraev
2015-03-30 20:15     ` Neil Horman

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='CAJ0CJ8nR2DFe32ULB_J+c7aY_mBXxJnubkHYwX_=rwFRnE7NqQ@mail.gmail.com' \
    --to=vadim.suraev@gmail.com \
    --cc=dev@dpdk.org \
    --cc=konstantin.ananyev@intel.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

DPDK patches and discussions

This inbox may be cloned and mirrored by anyone:

	git clone --mirror https://inbox.dpdk.org/dev/0 dev/git/0.git

	# If you have public-inbox 1.1+ installed, you may
	# initialize and index your mirror using the following commands:
	public-inbox-init -V2 dev dev/ https://inbox.dpdk.org/dev \
		dev@dpdk.org
	public-inbox-index dev

Example config snippet for mirrors.
Newsgroup available over NNTP:
	nntp://inbox.dpdk.org/inbox.dpdk.dev


AGPL code for this site: git clone https://public-inbox.org/public-inbox.git