From patchwork Thu Nov 23 01:45:37 2023 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Li RongQing X-Patchwork-Id: 13465682 X-Patchwork-Delegate: kuba@kernel.org Authentication-Results: smtp.subspace.kernel.org; dkim=none Received: from njjs-sys-mailin01.njjs.baidu.com (mx310.baidu.com [180.101.52.44]) by lindbergh.monkeyblade.net (Postfix) with ESMTP id 3109812A; Wed, 22 Nov 2023 17:45:39 -0800 (PST) Received: from localhost (bjhw-sys-rpm015653cc5.bjhw.baidu.com [10.227.53.39]) by njjs-sys-mailin01.njjs.baidu.com (Postfix) with ESMTP id E01AB7F000A9; Thu, 23 Nov 2023 09:45:37 +0800 (CST) From: Li RongQing To: linux-s390@vger.kernel.org, netdev@vger.kernel.org, wintera@linux.ibm.com, dust.li@linux.alibaba.com Subject: [PATCH net-next v4] net/smc: remove unneeded atomic operations in smc_tx_sndbuf_nonempty Date: Thu, 23 Nov 2023 09:45:37 +0800 Message-Id: <20231123014537.9786-1-lirongqing@baidu.com> X-Mailer: git-send-email 2.9.4 Precedence: bulk X-Mailing-List: netdev@vger.kernel.org List-Id: List-Subscribe: List-Unsubscribe: X-Patchwork-Delegate: kuba@kernel.org The commit dcd2cf5f2fc0 ("net/smc: add autocorking support") adds an atomic variable tx_pushing in smc_connection to make sure only one can send to let it cork more and save CDC slot. since smc_tx_pending can be called in the soft IRQ without checking sock_owned_by_user() at that time, which would cause a race condition because bh_lock_sock() did not honor sock_lock() After commit 6b88af839d20 ("net/smc: don't send in the BH context if sock_owned_by_user"), the transmission is deferred to when sock_lock() is held by the user. Therefore, we no longer need tx_pending to hold message. So remove atomic variable tx_pushing and its operation, and smc_tx_sndbuf_nonempty becomes a wrapper of __smc_tx_sndbuf_nonempty, so rename __smc_tx_sndbuf_nonempty back to smc_tx_sndbuf_nonempty Suggested-by: Alexandra Winter Co-developed-by: Dust Li Signed-off-by: Dust Li Signed-off-by: Li RongQing Reviewed-by: Alexandra Winter diff v4: remove atomic variable tx_pushing diff v3: improvements in the commit body and comments diff v2: fix a typo in commit body and add net-next subject-prefix net/smc/smc.h | 1 - net/smc/smc_tx.c | 30 +----------------------------- 2 files changed, 1 insertion(+), 30 deletions(-) diff --git a/net/smc/smc.h b/net/smc/smc.h index e377980..cd51261 100644 --- a/net/smc/smc.h +++ b/net/smc/smc.h @@ -196,7 +196,6 @@ struct smc_connection { * - dec on polled tx cqe */ wait_queue_head_t cdc_pend_tx_wq; /* wakeup on no cdc_pend_tx_wr*/ - atomic_t tx_pushing; /* nr_threads trying tx push */ struct delayed_work tx_work; /* retry of smc_cdc_msg_send */ u32 tx_off; /* base offset in peer rmb */ diff --git a/net/smc/smc_tx.c b/net/smc/smc_tx.c index 3b0ff3b..214ac3c 100644 --- a/net/smc/smc_tx.c +++ b/net/smc/smc_tx.c @@ -621,7 +621,7 @@ static int smcd_tx_sndbuf_nonempty(struct smc_connection *conn) return rc; } -static int __smc_tx_sndbuf_nonempty(struct smc_connection *conn) +int smc_tx_sndbuf_nonempty(struct smc_connection *conn) { struct smc_sock *smc = container_of(conn, struct smc_sock, conn); int rc = 0; @@ -655,34 +655,6 @@ static int __smc_tx_sndbuf_nonempty(struct smc_connection *conn) return rc; } -int smc_tx_sndbuf_nonempty(struct smc_connection *conn) -{ - int rc; - - /* This make sure only one can send simultaneously to prevent wasting - * of CPU and CDC slot. - * Record whether someone has tried to push while we are pushing. - */ - if (atomic_inc_return(&conn->tx_pushing) > 1) - return 0; - -again: - atomic_set(&conn->tx_pushing, 1); - smp_wmb(); /* Make sure tx_pushing is 1 before real send */ - rc = __smc_tx_sndbuf_nonempty(conn); - - /* We need to check whether someone else have added some data into - * the send queue and tried to push but failed after the atomic_set() - * when we are pushing. - * If so, we need to push again to prevent those data hang in the send - * queue. - */ - if (unlikely(!atomic_dec_and_test(&conn->tx_pushing))) - goto again; - - return rc; -} - /* Wakeup sndbuf consumers from process context * since there is more data to transmit. The caller * must hold sock lock.