From mboxrd@z Thu Jan 1 00:00:00 1970 From: Olivier MATZ Subject: Re: [PATCH v7 2/5] mempool: remove rte_ring from rte_mempool struct Date: Fri, 3 Jun 2016 14:28:45 +0200 Message-ID: <575177FD.4020908@6wind.com> References: <1464797998-76690-1-git-send-email-david.hunt@intel.com> <1464874043-67467-1-git-send-email-david.hunt@intel.com> <1464874043-67467-3-git-send-email-david.hunt@intel.com> Mime-Version: 1.0 Content-Type: text/plain; charset=windows-1252 Content-Transfer-Encoding: 7bit Cc: viktorin@rehivetech.com, jerin.jacob@caviumnetworks.com To: David Hunt , dev@dpdk.org Return-path: Received: from mail-wm0-f49.google.com (mail-wm0-f49.google.com [74.125.82.49]) by dpdk.org (Postfix) with ESMTP id 55B115A57 for ; Fri, 3 Jun 2016 14:28:48 +0200 (CEST) Received: by mail-wm0-f49.google.com with SMTP id a20so104254720wma.1 for ; Fri, 03 Jun 2016 05:28:48 -0700 (PDT) In-Reply-To: <1464874043-67467-3-git-send-email-david.hunt@intel.com> List-Id: patches and discussions about DPDK List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" On 06/02/2016 03:27 PM, David Hunt wrote: > Now that we're moving to an external mempoool handler, which > uses a void *pool_data as a pointer to the pool data, remove the > unneeded ring pointer from the mempool struct. > > Signed-off-by: David Hunt > --- > app/test/test_mempool_perf.c | 1 - > lib/librte_mempool/rte_mempool.h | 1 - > 2 files changed, 2 deletions(-) > > diff --git a/app/test/test_mempool_perf.c b/app/test/test_mempool_perf.c > index cdc02a0..091c1df 100644 > --- a/app/test/test_mempool_perf.c > +++ b/app/test/test_mempool_perf.c > @@ -161,7 +161,6 @@ per_lcore_mempool_test(__attribute__((unused)) void *arg) > n_get_bulk); > if (unlikely(ret < 0)) { > rte_mempool_dump(stdout, mp); > - rte_ring_dump(stdout, mp->ring); > /* in this case, objects are lost... */ > return -1; > } > diff --git a/lib/librte_mempool/rte_mempool.h b/lib/librte_mempool/rte_mempool.h > index a6b28b0..c33eeb8 100644 > --- a/lib/librte_mempool/rte_mempool.h > +++ b/lib/librte_mempool/rte_mempool.h > @@ -204,7 +204,6 @@ struct rte_mempool_memhdr { > */ > struct rte_mempool { > char name[RTE_MEMPOOL_NAMESIZE]; /**< Name of mempool. */ > - struct rte_ring *ring; /**< Ring to store objects. */ > union { > void *pool_data; /**< Ring or pool to store objects */ > uint64_t pool_id; /**< External mempool identifier */ > Sorry if I missed it in previous discussions, but I don't really see the point of having this in a separate commit, as the goal of the previous commit is to replace the ring by configurable ops. Moreover, after applying only the previous commit, the call to rte_ring_dump(stdout, mp->ring) would probably crash sine ring is NULL. I think this comment also applies to the next commit. Splitting between functionalities is good, but in this case I think the 3 commits are linked together, and it should not break compilation or tests to facilitate the git bisect. Regards, Olivier