From patchwork Thu Sep 6 17:12:14 2018 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Honnappa Nagarahalli X-Patchwork-Id: 44349 Return-Path: X-Original-To: patchwork@dpdk.org Delivered-To: patchwork@dpdk.org Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id A520D326D; Thu, 6 Sep 2018 19:12:28 +0200 (CEST) Received: from foss.arm.com (foss.arm.com [217.140.101.70]) by dpdk.org (Postfix) with ESMTP id 9C6EE326C for ; Thu, 6 Sep 2018 19:12:27 +0200 (CEST) Received: from usa-sjc-imap-foss1.foss.arm.com (unknown [10.72.51.249]) by usa-sjc-mx-foss1.foss.arm.com (Postfix) with ESMTP id 8A7397A9; Thu, 6 Sep 2018 10:12:26 -0700 (PDT) Received: from 2p2660v4-1.austin.arm.com (2p2660v4-1.austin.arm.com [10.118.12.164]) by usa-sjc-imap-foss1.foss.arm.com (Postfix) with ESMTPSA id 1CA2B3F557; Thu, 6 Sep 2018 10:12:26 -0700 (PDT) From: Honnappa Nagarahalli To: bruce.richardson@intel.com, pablo.de.lara.guarch@intel.com Cc: dev@dpdk.org, honnappa.nagarahalli@dpdk.org, gavin.hu@arm.com, steve.capper@arm.com, ola.liljedahl@arm.com, nd@arm.com, Honnappa Nagarahalli Date: Thu, 6 Sep 2018 12:12:14 -0500 Message-Id: <1536253938-192391-1-git-send-email-honnappa.nagarahalli@arm.com> X-Mailer: git-send-email 2.7.4 Subject: [dpdk-dev] [PATCH 0/4] Address reader-writer concurrency in rte_hash X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: dev-bounces@dpdk.org Sender: "dev" Currently, reader-writer concurrency problems in rte_hash are addressed using reader-writer locks. Use of reader-writer locks results in following issues: 1) In many of the use cases for the hash table, writer threads are running on control plane. If the writer is preempted while holding the lock, it will block the readers for an extended period resulting in packet drops. This problem seems to apply for platforms with transactional memory support as well because of the algorithm used for rte_rwlock_write_lock_tm: static inline void rte_rwlock_write_lock_tm(rte_rwlock_t *rwl) { if (likely(rte_try_tm(&rwl->cnt))) return; rte_rwlock_write_lock(rwl); } i.e. there is a posibility of using rte_rwlock_write_lock in failure cases. 2) Reader-writer lock based solution does not address the following issue. rte_hash_lookup_xxx APIs return the index of the element in the key store. Application(reader) can use that index to reference other data structures in its scope. Because of this, the index should not be freed till the application completes using the index. 3) Since writer blocks all the readers, the hash lookup rate comes down significantly when there is activity on the writer. This happens even for unrelated entries. Performance numbers given below clearly indicate this. Lock-free solution is required to solve these problems. This patch series adds the lock-free capabilities in the following steps: 1) Correct the alignment for the key store entry to prep for memory ordering. 2) Add memory ordering to prevent race conditions when a new key is added to the table. 3) Reader-writer concurrency issue, caused by moving the keys to their alternate locations during key insert, is solved by introducing an atomic global counter indicating a change in table. 4) This solution also has to solve the issue of readers using key store element even after the key is deleted from control plane. To solve this issue, the hash_del_key_xxx APIs do not free the key store element. The key store element has to be freed using the newly introduced rte_hash_free_key_with_position API. It needs to be called once all the readers have stopped using the key store element. How this is determined is outside the scope of this patch (RCU is one such mechanism that the application can use). 4) Finally, a lock free reader-writer concurrency flag is added to enable this feature at run time. Performance numbers: Scenario: Equal number of writer/reader threads for concurrent writers and readers. For readers only test, the entries are added upfront. Current code: Cores Lookup Lookup with add 2 474 246 4 935 579 6 1387 1048 8 1766 1480 10 2119 1951 12 2546 2441 With this patch: Cores Lookup Lookup with add 2 291 211 4 297 196 6 304 198 8 309 202 10 315 205 12 319 209 Honnappa Nagarahalli (4): hash: correct key store element alignment hash: add memory ordering to avoid race conditions hash: fix rw concurrency while moving keys hash: enable lock-free reader-writer concurrency lib/librte_hash/rte_cuckoo_hash.c | 445 +++++++++++++++++++++++++---------- lib/librte_hash/rte_cuckoo_hash.h | 6 +- lib/librte_hash/rte_hash.h | 63 ++++- lib/librte_hash/rte_hash_version.map | 7 + 4 files changed, 393 insertions(+), 128 deletions(-)