From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from dpdk.org (dpdk.org [92.243.14.124]) by inbox.dpdk.org (Postfix) with ESMTP id 66523A04B2; Wed, 26 Aug 2020 11:56:21 +0200 (CEST) Received: from [92.243.14.124] (localhost [127.0.0.1]) by dpdk.org (Postfix) with ESMTP id 4D2EB1BFE3; Wed, 26 Aug 2020 11:56:08 +0200 (CEST) Received: from mga09.intel.com (mga09.intel.com [134.134.136.24]) by dpdk.org (Postfix) with ESMTP id 5B1C71BFE3 for ; Wed, 26 Aug 2020 11:56:06 +0200 (CEST) IronPort-SDR: /WLN4JbGD/u93hp1sNTs6jQ01LpTdIgFJakmNfs/KSgK8yKcAXzabc6+l57CPRbq7GxGroAbu8 HhBDNp+5fbMQ== X-IronPort-AV: E=McAfee;i="6000,8403,9724"; a="157301345" X-IronPort-AV: E=Sophos;i="5.76,355,1592895600"; d="scan'208";a="157301345" X-Amp-Result: SKIPPED(no attachment in message) X-Amp-File-Uploaded: False Received: from orsmga003.jf.intel.com ([10.7.209.27]) by orsmga102.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 26 Aug 2020 02:56:06 -0700 IronPort-SDR: q2aoME4Fa47f1+b0QUAfLn+vmZsS19ExCbJSJ6FDxqVs4Gf6mIyRdIUd437KsT0pcnCQ8GYPdP 3kACHwdpcuEg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.76,355,1592895600"; d="scan'208";a="295319490" Received: from silpixa00400567.ir.intel.com ([10.237.214.190]) by orsmga003.jf.intel.com with ESMTP; 26 Aug 2020 02:56:03 -0700 From: Radu Nicolau To: dev@dpdk.org Cc: beilei.xing@intel.com, jia.guo@intel.com, bruce.richardson@intel.com, konstantin.ananyev@intel.com, jerinjacobk@gmail.com, david.marchand@redhat.com, fiona.trahe@intel.com, wei.zhao1@intel.com, ruifeng.wang@arm.com, qiming.yang@intel.com, qi.z.zhang@intel.com, Radu Nicolau Date: Wed, 26 Aug 2020 10:55:49 +0100 Message-Id: <20200826095552.82525-3-radu.nicolau@intel.com> X-Mailer: git-send-email 2.17.1 In-Reply-To: <20200826095552.82525-1-radu.nicolau@intel.com> References: <1591870283-7776-1-git-send-email-radu.nicolau@intel.com> <20200826095552.82525-1-radu.nicolau@intel.com> Subject: [dpdk-dev] [PATCH v11 2/5] net/i40e: use WC store to update queue tail registers 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" Performance improvement: use a write combining store instead of a regular mmio write to update queue tail registers. Signed-off-by: Radu Nicolau Acked-by: Bruce Richardson --- drivers/net/i40e/base/i40e_osdep.h | 5 +++++ drivers/net/i40e/i40e_rxtx.c | 8 ++++---- drivers/net/i40e/i40e_rxtx_vec_avx2.c | 4 ++-- drivers/net/i40e/i40e_rxtx_vec_sse.c | 4 ++-- 4 files changed, 13 insertions(+), 8 deletions(-) diff --git a/drivers/net/i40e/base/i40e_osdep.h b/drivers/net/i40e/base/i40e_osdep.h index 58be39677..9b5033024 100644 --- a/drivers/net/i40e/base/i40e_osdep.h +++ b/drivers/net/i40e/base/i40e_osdep.h @@ -138,6 +138,11 @@ static inline uint32_t i40e_read_addr(volatile void *addr) #define I40E_PCI_REG_WRITE_RELAXED(reg, value) \ rte_write32_relaxed((rte_cpu_to_le_32(value)), reg) +#define I40E_PCI_REG_WC_WRITE(reg, value) \ + rte_write32_wc((rte_cpu_to_le_32(value)), reg) +#define I40E_PCI_REG_WC_WRITE_RELAXED(reg, value) \ + rte_write32_wc_relaxed((rte_cpu_to_le_32(value)), reg) + #define I40E_WRITE_FLUSH(a) I40E_READ_REG(a, I40E_GLGEN_STAT) #define I40EVF_WRITE_FLUSH(a) I40E_READ_REG(a, I40E_VFGEN_RSTAT) diff --git a/drivers/net/i40e/i40e_rxtx.c b/drivers/net/i40e/i40e_rxtx.c index fe7f9200c..76d634091 100644 --- a/drivers/net/i40e/i40e_rxtx.c +++ b/drivers/net/i40e/i40e_rxtx.c @@ -760,7 +760,7 @@ i40e_recv_pkts(void *rx_queue, struct rte_mbuf **rx_pkts, uint16_t nb_pkts) if (nb_hold > rxq->rx_free_thresh) { rx_id = (uint16_t) ((rx_id == 0) ? (rxq->nb_rx_desc - 1) : (rx_id - 1)); - I40E_PCI_REG_WRITE(rxq->qrx_tail, rx_id); + I40E_PCI_REG_WC_WRITE(rxq->qrx_tail, rx_id); nb_hold = 0; } rxq->nb_rx_hold = nb_hold; @@ -938,7 +938,7 @@ i40e_recv_scattered_pkts(void *rx_queue, if (nb_hold > rxq->rx_free_thresh) { rx_id = (uint16_t)(rx_id == 0 ? (rxq->nb_rx_desc - 1) : (rx_id - 1)); - I40E_PCI_REG_WRITE(rxq->qrx_tail, rx_id); + I40E_PCI_REG_WC_WRITE(rxq->qrx_tail, rx_id); nb_hold = 0; } rxq->nb_rx_hold = nb_hold; @@ -1249,7 +1249,7 @@ i40e_xmit_pkts(void *tx_queue, struct rte_mbuf **tx_pkts, uint16_t nb_pkts) (unsigned) tx_id, (unsigned) nb_tx); rte_cio_wmb(); - I40E_PCI_REG_WRITE_RELAXED(txq->qtx_tail, tx_id); + I40E_PCI_REG_WC_WRITE_RELAXED(txq->qtx_tail, tx_id); txq->tx_tail = tx_id; return nb_tx; @@ -1400,7 +1400,7 @@ tx_xmit_pkts(struct i40e_tx_queue *txq, txq->tx_tail = 0; /* Update the tx tail register */ - I40E_PCI_REG_WRITE(txq->qtx_tail, txq->tx_tail); + I40E_PCI_REG_WC_WRITE(txq->qtx_tail, txq->tx_tail); return nb_pkts; } diff --git a/drivers/net/i40e/i40e_rxtx_vec_avx2.c b/drivers/net/i40e/i40e_rxtx_vec_avx2.c index 3bcef1363..178d8f4e2 100644 --- a/drivers/net/i40e/i40e_rxtx_vec_avx2.c +++ b/drivers/net/i40e/i40e_rxtx_vec_avx2.c @@ -134,7 +134,7 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) (rxq->nb_rx_desc - 1) : (rxq->rxrearm_start - 1)); /* Update the tail pointer on the NIC */ - I40E_PCI_REG_WRITE(rxq->qrx_tail, rx_id); + I40E_PCI_REG_WC_WRITE(rxq->qrx_tail, rx_id); } #ifndef RTE_LIBRTE_I40E_16BYTE_RX_DESC @@ -921,7 +921,7 @@ i40e_xmit_fixed_burst_vec_avx2(void *tx_queue, struct rte_mbuf **tx_pkts, txq->tx_tail = tx_id; - I40E_PCI_REG_WRITE(txq->qtx_tail, txq->tx_tail); + I40E_PCI_REG_WC_WRITE(txq->qtx_tail, txq->tx_tail); return nb_pkts; } diff --git a/drivers/net/i40e/i40e_rxtx_vec_sse.c b/drivers/net/i40e/i40e_rxtx_vec_sse.c index 698518349..240ce478a 100644 --- a/drivers/net/i40e/i40e_rxtx_vec_sse.c +++ b/drivers/net/i40e/i40e_rxtx_vec_sse.c @@ -86,7 +86,7 @@ i40e_rxq_rearm(struct i40e_rx_queue *rxq) (rxq->nb_rx_desc - 1) : (rxq->rxrearm_start - 1)); /* Update the tail pointer on the NIC */ - I40E_PCI_REG_WRITE(rxq->qrx_tail, rx_id); + I40E_PCI_REG_WC_WRITE(rxq->qrx_tail, rx_id); } #ifndef RTE_LIBRTE_I40E_16BYTE_RX_DESC @@ -733,7 +733,7 @@ i40e_xmit_fixed_burst_vec(void *tx_queue, struct rte_mbuf **tx_pkts, txq->tx_tail = tx_id; - I40E_PCI_REG_WRITE(txq->qtx_tail, txq->tx_tail); + I40E_PCI_REG_WC_WRITE(txq->qtx_tail, txq->tx_tail); return nb_pkts; } -- 2.17.1