From patchwork Thu Apr 1 02:11:12 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176865 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 96C14C43462 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 6ACA4610A8 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S229497AbhDACIA (ORCPT ); Wed, 31 Mar 2021 22:08:00 -0400 Received: from mga01.intel.com ([192.55.52.88]:60202 "EHLO mga01.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232763AbhDACHk (ORCPT ); Wed, 31 Mar 2021 22:07:40 -0400 IronPort-SDR: JI0WbZMMaT8qjAHFVCxm0w2NjIfhfDfnQjOmKR6voDQM+8Gao5tK/Fc37ZMuId5M7lQaIVhKYD fmU+m1KfQq1w== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="212379990" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="212379990" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by fmsmga101.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:07:40 -0700 IronPort-SDR: crsTts9aLDE4bskpm+aZ6m/kptAmCNFNvl5HOc6p3fhGMKCM6OGoygriVMZHQGhsmz0ZCg5y59 IiBH9hQjmTvg== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004198" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:07:35 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 1/6] net: stmmac: set IRQ affinity hint for multi MSI vectors Date: Thu, 1 Apr 2021 10:11:12 +0800 Message-Id: <20210401021117.13360-2-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org Certain platform likes Intel mGBE has independent hardware IRQ resources for TX and RX DMA operation. In preparation to support XDP TX, we add IRQ affinity hint to group both RX and TX queue of the same queue ID to the same CPU. Changes in v2: - IRQ affinity hint need to set to null before IRQ is released. Thanks to issue reported by Song, Yoong Siang. Reported-by: Song, Yoong Siang Signed-off-by: Ong Boon Leong --- drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 15 +++++++++++++-- 1 file changed, 13 insertions(+), 2 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index d34388b1ffcc..9d63e8c365ae 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -3005,15 +3005,19 @@ static void stmmac_free_irq(struct net_device *dev, fallthrough; case REQ_IRQ_ERR_TX: for (j = irq_idx - 1; j >= 0; j--) { - if (priv->tx_irq[j] > 0) + if (priv->tx_irq[j] > 0) { + irq_set_affinity_hint(priv->tx_irq[j], NULL); free_irq(priv->tx_irq[j], &priv->tx_queue[j]); + } } irq_idx = priv->plat->rx_queues_to_use; fallthrough; case REQ_IRQ_ERR_RX: for (j = irq_idx - 1; j >= 0; j--) { - if (priv->rx_irq[j] > 0) + if (priv->rx_irq[j] > 0) { + irq_set_affinity_hint(priv->rx_irq[j], NULL); free_irq(priv->rx_irq[j], &priv->rx_queue[j]); + } } if (priv->sfty_ue_irq > 0 && priv->sfty_ue_irq != dev->irq) @@ -3045,6 +3049,7 @@ static int stmmac_request_irq_multi_msi(struct net_device *dev) { enum request_irq_err irq_err = REQ_IRQ_ERR_NO; struct stmmac_priv *priv = netdev_priv(dev); + cpumask_t cpu_mask; int irq_idx = 0; char *int_name; int ret; @@ -3153,6 +3158,9 @@ static int stmmac_request_irq_multi_msi(struct net_device *dev) irq_idx = i; goto irq_error; } + cpumask_clear(&cpu_mask); + cpumask_set_cpu(i % num_online_cpus(), &cpu_mask); + irq_set_affinity_hint(priv->rx_irq[i], &cpu_mask); } /* Request Tx MSI irq */ @@ -3173,6 +3181,9 @@ static int stmmac_request_irq_multi_msi(struct net_device *dev) irq_idx = i; goto irq_error; } + cpumask_clear(&cpu_mask); + cpumask_set_cpu(i % num_online_cpus(), &cpu_mask); + irq_set_affinity_hint(priv->tx_irq[i], &cpu_mask); } return 0; From patchwork Thu Apr 1 02:11:13 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176867 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id A867EC43470 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 8224B610CB for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S233090AbhDACIA (ORCPT ); Wed, 31 Mar 2021 22:08:00 -0400 Received: from mga12.intel.com ([192.55.52.136]:22549 "EHLO mga12.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232967AbhDACHq (ORCPT ); Wed, 31 Mar 2021 22:07:46 -0400 IronPort-SDR: 41gh7r48m1i1ovNYiCo3ZYaCV51kZ/ock76sneXf5hySJYukB+hDHoboSjPnzT5r+8t8o7Wc5a 5WBL1QnlPoFw== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="171547798" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="171547798" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by fmsmga106.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:07:46 -0700 IronPort-SDR: nbGoBPcLLGTjmaCEl0zzbCgZoGVn1aKC9AVezmMzl1gtBtCGJTaz/QIQaOEVYI1330cvG7qhvQ I3oIQJrSr+lw== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004218" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:07:40 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 2/6] net: stmmac: make SPH enable/disable to be configurable Date: Thu, 1 Apr 2021 10:11:13 +0800 Message-Id: <20210401021117.13360-3-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org SPH functionality splits header and payload according to split mode and offsef fields (SPLM and SPLOFST). It is beneficials for Linux network stack RX processing however it adds a lot of complexity in XDP processing. So, this patch makes the split-header (SPH) capability of the controller is stored in "priv->sph_cap" and the enabling/disabling of SPH is decided by "priv->sph". This is to prepare initial XDP enabling for stmmac to disable the use of SPH whenever XDP is enabled. Signed-off-by: Ong Boon Leong --- drivers/net/ethernet/stmicro/stmmac/stmmac.h | 1 + drivers/net/ethernet/stmicro/stmmac/stmmac_main.c | 12 +++++++----- 2 files changed, 8 insertions(+), 5 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h index 9966f6f10905..e293423f98c3 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h @@ -160,6 +160,7 @@ struct stmmac_priv { bool tx_path_in_lpi_mode; bool tso; int sph; + int sph_cap; u32 sarc_type; unsigned int dma_buf_sz; diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 9d63e8c365ae..18e34a1e2367 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -2858,6 +2858,7 @@ static int stmmac_hw_setup(struct net_device *dev, bool init_ptp) struct stmmac_priv *priv = netdev_priv(dev); u32 rx_cnt = priv->plat->rx_queues_to_use; u32 tx_cnt = priv->plat->tx_queues_to_use; + bool sph_en; u32 chan; int ret; @@ -2952,10 +2953,10 @@ static int stmmac_hw_setup(struct net_device *dev, bool init_ptp) } /* Enable Split Header */ - if (priv->sph && priv->hw->rx_csum) { - for (chan = 0; chan < rx_cnt; chan++) - stmmac_enable_sph(priv, priv->ioaddr, 1, chan); - } + sph_en = (priv->hw->rx_csum > 0) && priv->sph; + for (chan = 0; chan < rx_cnt; chan++) + stmmac_enable_sph(priv, priv->ioaddr, sph_en, chan); + /* VLAN Tag Insertion */ if (priv->dma_cap.vlins) @@ -5708,7 +5709,8 @@ int stmmac_dvr_probe(struct device *device, if (priv->dma_cap.sphen) { ndev->hw_features |= NETIF_F_GRO; - priv->sph = true; + priv->sph_cap = true; + priv->sph = priv->sph_cap; dev_info(priv->device, "SPH feature enabled\n"); } From patchwork Thu Apr 1 02:11:14 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 8bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176869 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id B223AC43603 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 93C1A610CD for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S233125AbhDACIB (ORCPT ); Wed, 31 Mar 2021 22:08:01 -0400 Received: from mga18.intel.com ([134.134.136.126]:12618 "EHLO mga18.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S232994AbhDACHw (ORCPT ); Wed, 31 Mar 2021 22:07:52 -0400 IronPort-SDR: LGbWlgLKDWw4Hqw6pDgm/nsTrzSnO9tJ2pGgAKnZ/PeNPemSrwz/K2v9SD1Abwmbl2qpN9AALk He6/Ew99UyAA== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="179664031" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="179664031" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by orsmga106.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:07:52 -0700 IronPort-SDR: PvCDMZ9NQ+z+wHv91b6TMXnArKvUx95GfkH3AZtNyR8jA2abhgospZRbbnWHgpChrFr8SEYuSG I2DVvL45C1oQ== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004231" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:07:46 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 3/6] net: stmmac: arrange Tx tail pointer update to stmmac_flush_tx_descriptors Date: Thu, 1 Apr 2021 10:11:14 +0800 Message-Id: <20210401021117.13360-4-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org This patch organizes TX tail pointer update into a new function called stmmac_flush_tx_descriptors() so that we can reuse it in stmmac_xmit(), stmmac_tso_xmit() and up-coming XDP implementation. Changes to v2: - Fix for warning: unused variable ‘desc_size’ https://patchwork.hopto.org/static/nipa/457321/12170149/build_32bit/stderr Signed-off-by: Ong Boon Leong --- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 56 +++++++++---------- 1 file changed, 26 insertions(+), 30 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 18e34a1e2367..cb1b2a180429 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -3518,6 +3518,28 @@ static void stmmac_tso_allocator(struct stmmac_priv *priv, dma_addr_t des, } } +static void stmmac_flush_tx_descriptors(struct stmmac_priv *priv, int queue) +{ + struct stmmac_tx_queue *tx_q = &priv->tx_queue[queue]; + int desc_size; + + if (likely(priv->extend_desc)) + desc_size = sizeof(struct dma_extended_desc); + else if (tx_q->tbs & STMMAC_TBS_AVAIL) + desc_size = sizeof(struct dma_edesc); + else + desc_size = sizeof(struct dma_desc); + + /* The own bit must be the latest setting done when prepare the + * descriptor and then barrier is needed to make sure that + * all is coherent before granting the DMA engine. + */ + wmb(); + + tx_q->tx_tail_addr = tx_q->dma_tx_phy + (tx_q->cur_tx * desc_size); + stmmac_set_tx_tail_ptr(priv, priv->ioaddr, tx_q->tx_tail_addr, queue); +} + /** * stmmac_tso_xmit - Tx entry point of the driver for oversized frames (TSO) * @skb : the socket buffer @@ -3549,10 +3571,10 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev) { struct dma_desc *desc, *first, *mss_desc = NULL; struct stmmac_priv *priv = netdev_priv(dev); - int desc_size, tmp_pay_len = 0, first_tx; int nfrags = skb_shinfo(skb)->nr_frags; u32 queue = skb_get_queue_mapping(skb); unsigned int first_entry, tx_packets; + int tmp_pay_len = 0, first_tx; struct stmmac_tx_queue *tx_q; bool has_vlan, set_ic; u8 proto_hdr_len, hdr; @@ -3750,12 +3772,6 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev) stmmac_set_tx_owner(priv, mss_desc); } - /* The own bit must be the latest setting done when prepare the - * descriptor and then barrier is needed to make sure that - * all is coherent before granting the DMA engine. - */ - wmb(); - if (netif_msg_pktdata(priv)) { pr_info("%s: curr=%d dirty=%d f=%d, e=%d, f_p=%p, nfrags %d\n", __func__, tx_q->cur_tx, tx_q->dirty_tx, first_entry, @@ -3766,13 +3782,7 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev) netdev_tx_sent_queue(netdev_get_tx_queue(dev, queue), skb->len); - if (tx_q->tbs & STMMAC_TBS_AVAIL) - desc_size = sizeof(struct dma_edesc); - else - desc_size = sizeof(struct dma_desc); - - tx_q->tx_tail_addr = tx_q->dma_tx_phy + (tx_q->cur_tx * desc_size); - stmmac_set_tx_tail_ptr(priv, priv->ioaddr, tx_q->tx_tail_addr, queue); + stmmac_flush_tx_descriptors(priv, queue); stmmac_tx_timer_arm(priv, queue); return NETDEV_TX_OK; @@ -3802,10 +3812,10 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev) int nfrags = skb_shinfo(skb)->nr_frags; int gso = skb_shinfo(skb)->gso_type; struct dma_edesc *tbs_desc = NULL; - int entry, desc_size, first_tx; struct dma_desc *desc, *first; struct stmmac_tx_queue *tx_q; bool has_vlan, set_ic; + int entry, first_tx; dma_addr_t des; tx_q = &priv->tx_queue[queue]; @@ -4007,25 +4017,11 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev) stmmac_set_tx_owner(priv, first); - /* The own bit must be the latest setting done when prepare the - * descriptor and then barrier is needed to make sure that - * all is coherent before granting the DMA engine. - */ - wmb(); - netdev_tx_sent_queue(netdev_get_tx_queue(dev, queue), skb->len); stmmac_enable_dma_transmission(priv, priv->ioaddr); - if (likely(priv->extend_desc)) - desc_size = sizeof(struct dma_extended_desc); - else if (tx_q->tbs & STMMAC_TBS_AVAIL) - desc_size = sizeof(struct dma_edesc); - else - desc_size = sizeof(struct dma_desc); - - tx_q->tx_tail_addr = tx_q->dma_tx_phy + (tx_q->cur_tx * desc_size); - stmmac_set_tx_tail_ptr(priv, priv->ioaddr, tx_q->tx_tail_addr, queue); + stmmac_flush_tx_descriptors(priv, queue); stmmac_tx_timer_arm(priv, queue); return NETDEV_TX_OK; From patchwork Thu Apr 1 02:11:15 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176871 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id CD7E0C43600 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id B761F61077 for ; Thu, 1 Apr 2021 02:08:33 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S233129AbhDACIC (ORCPT ); Wed, 31 Mar 2021 22:08:02 -0400 Received: from mga18.intel.com ([134.134.136.126]:12618 "EHLO mga18.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S233028AbhDACH5 (ORCPT ); Wed, 31 Mar 2021 22:07:57 -0400 IronPort-SDR: d+r7RaSawSR7WIVxF7cGLC95tU19ana+NlimxT9MUefeBI8cdDN1WB8O6dhDm6Im6FbVMDrpOR YTBp9+OS69zA== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="179664039" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="179664039" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by orsmga106.jf.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:07:57 -0700 IronPort-SDR: bcEWFjFuN35gsuu+J+7cu0jXSpXJR7LX3ZIRCvYDIExIbSjXFTxqIe9AJzLS0JSM6DgX77IUje XNHaZ9Y1BA/A== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004248" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:07:51 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 4/6] net: stmmac: Add initial XDP support Date: Thu, 1 Apr 2021 10:11:15 +0800 Message-Id: <20210401021117.13360-5-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org This patch adds the initial XDP support to stmmac driver. It supports XDP_PASS, XDP_DROP and XDP_ABORTED actions. Upcoming patches will add support for XDP_TX and XDP_REDIRECT. To support XDP headroom, this patch adds page_offset into RX buffer and change the dma_sync_single_for_device|cpu(). The DMA address used for RX operation are changed to take into page_offset too. As page_pool can handle dma_sync_single_for_device() on behalf of driver with PP_FLAG_DMA_SYNC_DEV flag, we skip doing that in stmmac driver. Current stmmac driver supports split header support (SPH) in RX but the flexibility of splitting header and payload at different position makes it very complex to be supported for XDP processing. In addition, jumbo frame is not supported in XDP to keep the initial codes simple. This patch has been tested with the sample app "xdp1" located in samples/bpf directory for both SKB and Native (XDP) mode. The burst traffic generated using pktgen_sample03_burst_single_flow.sh in samples/pktgen directory. Changes in v3: - factor in xdp header and tail adjustment done by XDP program. Thanks to Jakub Kicinski for pointing out the gap in v2. Changes in v2: - fix for "warning: variable 'len' set but not used" reported by lkp. Reported-by: kernel test robot Signed-off-by: Ong Boon Leong --- drivers/net/ethernet/stmicro/stmmac/Makefile | 1 + drivers/net/ethernet/stmicro/stmmac/stmmac.h | 21 ++- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 150 +++++++++++++++--- .../net/ethernet/stmicro/stmmac/stmmac_xdp.c | 40 +++++ .../net/ethernet/stmicro/stmmac/stmmac_xdp.h | 12 ++ 5 files changed, 197 insertions(+), 27 deletions(-) create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.c create mode 100644 drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.h diff --git a/drivers/net/ethernet/stmicro/stmmac/Makefile b/drivers/net/ethernet/stmicro/stmmac/Makefile index 366740ab9c5a..f2e478b884b0 100644 --- a/drivers/net/ethernet/stmicro/stmmac/Makefile +++ b/drivers/net/ethernet/stmicro/stmmac/Makefile @@ -6,6 +6,7 @@ stmmac-objs:= stmmac_main.o stmmac_ethtool.o stmmac_mdio.o ring_mode.o \ mmc_core.o stmmac_hwtstamp.o stmmac_ptp.o dwmac4_descs.o \ dwmac4_dma.o dwmac4_lib.o dwmac4_core.o dwmac5.o hwif.o \ stmmac_tc.o dwxgmac2_core.o dwxgmac2_dma.o dwxgmac2_descs.o \ + stmmac_xdp.o \ $(stmmac-y) stmmac-$(CONFIG_STMMAC_SELFTESTS) += stmmac_selftests.o diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h index e293423f98c3..e72224c8fbac 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h @@ -68,8 +68,9 @@ struct stmmac_tx_queue { struct stmmac_rx_buffer { struct page *page; - struct page *sec_page; dma_addr_t addr; + __u32 page_offset; + struct page *sec_page; dma_addr_t sec_addr; }; @@ -269,6 +270,9 @@ struct stmmac_priv { /* Receive Side Scaling */ struct stmmac_rss rss; + + /* XDP BPF Program */ + struct bpf_prog *xdp_prog; }; enum stmmac_state { @@ -285,6 +289,8 @@ void stmmac_set_ethtool_ops(struct net_device *netdev); void stmmac_ptp_register(struct stmmac_priv *priv); void stmmac_ptp_unregister(struct stmmac_priv *priv); +int stmmac_open(struct net_device *dev); +int stmmac_release(struct net_device *dev); int stmmac_resume(struct device *dev); int stmmac_suspend(struct device *dev); int stmmac_dvr_remove(struct device *dev); @@ -298,6 +304,19 @@ int stmmac_reinit_ringparam(struct net_device *dev, u32 rx_size, u32 tx_size); int stmmac_bus_clks_config(struct stmmac_priv *priv, bool enabled); void stmmac_fpe_handshake(struct stmmac_priv *priv, bool enable); +static inline bool stmmac_xdp_is_enabled(struct stmmac_priv *priv) +{ + return !!priv->xdp_prog; +} + +static inline unsigned int stmmac_rx_offset(struct stmmac_priv *priv) +{ + if (stmmac_xdp_is_enabled(priv)) + return XDP_PACKET_HEADROOM; + + return 0; +} + #if IS_ENABLED(CONFIG_STMMAC_SELFTESTS) void stmmac_selftest_run(struct net_device *dev, struct ethtool_test *etest, u64 *buf); diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index cb1b2a180429..0dad8ab93eb5 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -38,9 +38,11 @@ #include #include #include +#include #include #include "stmmac_ptp.h" #include "stmmac.h" +#include "stmmac_xdp.h" #include #include #include "dwmac1000.h" @@ -67,6 +69,9 @@ MODULE_PARM_DESC(phyaddr, "Physical device address"); #define STMMAC_TX_THRESH(x) ((x)->dma_tx_size / 4) #define STMMAC_RX_THRESH(x) ((x)->dma_rx_size / 4) +#define STMMAC_XDP_PASS 0 +#define STMMAC_XDP_CONSUMED BIT(0) + static int flow_ctrl = FLOW_AUTO; module_param(flow_ctrl, int, 0644); MODULE_PARM_DESC(flow_ctrl, "Flow control ability [on/off]"); @@ -1384,6 +1389,7 @@ static int stmmac_init_rx_buffers(struct stmmac_priv *priv, struct dma_desc *p, buf->page = page_pool_dev_alloc_pages(rx_q->page_pool); if (!buf->page) return -ENOMEM; + buf->page_offset = stmmac_rx_offset(priv); if (priv->sph) { buf->sec_page = page_pool_dev_alloc_pages(rx_q->page_pool); @@ -1397,7 +1403,8 @@ static int stmmac_init_rx_buffers(struct stmmac_priv *priv, struct dma_desc *p, stmmac_set_desc_sec_addr(priv, p, buf->sec_addr, false); } - buf->addr = page_pool_get_dma_addr(buf->page); + buf->addr = page_pool_get_dma_addr(buf->page) + buf->page_offset; + stmmac_set_desc_addr(priv, p, buf->addr); if (priv->dma_buf_sz == BUF_SIZE_16KiB) stmmac_init_desc3(priv, p); @@ -1503,7 +1510,8 @@ static void stmmac_reinit_rx_buffers(struct stmmac_priv *priv) if (!buf->page) goto err_reinit_rx_buffers; - buf->addr = page_pool_get_dma_addr(buf->page); + buf->addr = page_pool_get_dma_addr(buf->page) + + buf->page_offset; } if (priv->sph && !buf->sec_page) { @@ -1821,6 +1829,7 @@ static void free_dma_tx_desc_resources(struct stmmac_priv *priv) */ static int alloc_dma_rx_desc_resources(struct stmmac_priv *priv) { + bool xdp_prog = stmmac_xdp_is_enabled(priv); u32 rx_count = priv->plat->rx_queues_to_use; int ret = -ENOMEM; u32 queue; @@ -1834,13 +1843,15 @@ static int alloc_dma_rx_desc_resources(struct stmmac_priv *priv) rx_q->queue_index = queue; rx_q->priv_data = priv; - pp_params.flags = PP_FLAG_DMA_MAP; + pp_params.flags = PP_FLAG_DMA_MAP | PP_FLAG_DMA_SYNC_DEV; pp_params.pool_size = priv->dma_rx_size; num_pages = DIV_ROUND_UP(priv->dma_buf_sz, PAGE_SIZE); pp_params.order = ilog2(num_pages); pp_params.nid = dev_to_node(priv->device); pp_params.dev = priv->device; - pp_params.dma_dir = DMA_FROM_DEVICE; + pp_params.dma_dir = xdp_prog ? DMA_BIDIRECTIONAL : DMA_FROM_DEVICE; + pp_params.offset = stmmac_rx_offset(priv); + pp_params.max_len = STMMAC_MAX_RX_BUF_SIZE(num_pages); rx_q->page_pool = page_pool_create(&pp_params); if (IS_ERR(rx_q->page_pool)) { @@ -3268,7 +3279,7 @@ static int stmmac_request_irq(struct net_device *dev) * 0 on success and an appropriate (-)ve integer as defined in errno.h * file on failure. */ -static int stmmac_open(struct net_device *dev) +int stmmac_open(struct net_device *dev) { struct stmmac_priv *priv = netdev_priv(dev); int bfsize = 0; @@ -3391,7 +3402,7 @@ static void stmmac_fpe_stop_wq(struct stmmac_priv *priv) * Description: * This is the stop entry point of the driver. */ -static int stmmac_release(struct net_device *dev) +int stmmac_release(struct net_device *dev) { struct stmmac_priv *priv = netdev_priv(dev); u32 chan; @@ -4064,11 +4075,9 @@ static void stmmac_rx_vlan(struct net_device *dev, struct sk_buff *skb) static inline void stmmac_rx_refill(struct stmmac_priv *priv, u32 queue) { struct stmmac_rx_queue *rx_q = &priv->rx_queue[queue]; - int len, dirty = stmmac_rx_dirty(priv, queue); + int dirty = stmmac_rx_dirty(priv, queue); unsigned int entry = rx_q->dirty_rx; - len = DIV_ROUND_UP(priv->dma_buf_sz, PAGE_SIZE) * PAGE_SIZE; - while (dirty-- > 0) { struct stmmac_rx_buffer *buf = &rx_q->buf_pool[entry]; struct dma_desc *p; @@ -4091,18 +4100,9 @@ static inline void stmmac_rx_refill(struct stmmac_priv *priv, u32 queue) break; buf->sec_addr = page_pool_get_dma_addr(buf->sec_page); - - dma_sync_single_for_device(priv->device, buf->sec_addr, - len, DMA_FROM_DEVICE); } - buf->addr = page_pool_get_dma_addr(buf->page); - - /* Sync whole allocation to device. This will invalidate old - * data. - */ - dma_sync_single_for_device(priv->device, buf->addr, len, - DMA_FROM_DEVICE); + buf->addr = page_pool_get_dma_addr(buf->page) + buf->page_offset; stmmac_set_desc_addr(priv, p, buf->addr); if (priv->sph) @@ -4181,6 +4181,42 @@ static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, return plen - len; } +static struct sk_buff *stmmac_xdp_run_prog(struct stmmac_priv *priv, + struct xdp_buff *xdp) +{ + struct bpf_prog *prog; + int res; + u32 act; + + rcu_read_lock(); + + prog = READ_ONCE(priv->xdp_prog); + if (!prog) { + res = STMMAC_XDP_PASS; + goto unlock; + } + + act = bpf_prog_run_xdp(prog, xdp); + switch (act) { + case XDP_PASS: + res = STMMAC_XDP_PASS; + break; + default: + bpf_warn_invalid_xdp_action(act); + fallthrough; + case XDP_ABORTED: + trace_xdp_exception(priv->dev, prog, act); + fallthrough; + case XDP_DROP: + res = STMMAC_XDP_CONSUMED; + break; + } + +unlock: + rcu_read_unlock(); + return ERR_PTR(-res); +} + /** * stmmac_rx - manage the receive process * @priv: driver private structure @@ -4196,8 +4232,14 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) unsigned int count = 0, error = 0, len = 0; int status = 0, coe = priv->hw->rx_csum; unsigned int next_entry = rx_q->cur_rx; + enum dma_data_direction dma_dir; unsigned int desc_size; struct sk_buff *skb = NULL; + struct xdp_buff xdp; + int buf_sz; + + dma_dir = page_pool_get_dma_dir(rx_q->page_pool); + buf_sz = DIV_ROUND_UP(priv->dma_buf_sz, PAGE_SIZE) * PAGE_SIZE; if (netif_msg_rx_status(priv)) { void *rx_head; @@ -4315,6 +4357,45 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) } if (!skb) { + dma_sync_single_for_cpu(priv->device, buf->addr, + buf1_len, dma_dir); + + xdp.data = page_address(buf->page) + buf->page_offset; + xdp.data_end = xdp.data + buf1_len; + xdp.data_hard_start = page_address(buf->page); + xdp_set_data_meta_invalid(&xdp); + xdp.frame_sz = buf_sz; + + skb = stmmac_xdp_run_prog(priv, &xdp); + + /* For Not XDP_PASS verdict */ + if (IS_ERR(skb)) { + unsigned int xdp_res = -PTR_ERR(skb); + + if (xdp_res & STMMAC_XDP_CONSUMED) { + page_pool_recycle_direct(rx_q->page_pool, + buf->page); + buf->page = NULL; + priv->dev->stats.rx_dropped++; + + /* Clear skb as it was set as + * status by XDP program. + */ + skb = NULL; + + if (unlikely((status & rx_not_ls))) + goto read_again; + + count++; + continue; + } + } + } + + if (!skb) { + /* XDP program may expand or reduce tail */ + buf1_len = xdp.data_end - xdp.data; + skb = napi_alloc_skb(&ch->rx_napi, buf1_len); if (!skb) { priv->dev->stats.rx_dropped++; @@ -4322,10 +4403,8 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) goto drain_data; } - dma_sync_single_for_cpu(priv->device, buf->addr, - buf1_len, DMA_FROM_DEVICE); - skb_copy_to_linear_data(skb, page_address(buf->page), - buf1_len); + /* XDP program may adjust header */ + skb_copy_to_linear_data(skb, xdp.data, buf1_len); skb_put(skb, buf1_len); /* Data payload copied into SKB, page ready for recycle */ @@ -4333,9 +4412,9 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) buf->page = NULL; } else if (buf1_len) { dma_sync_single_for_cpu(priv->device, buf->addr, - buf1_len, DMA_FROM_DEVICE); + buf1_len, dma_dir); skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, - buf->page, 0, buf1_len, + buf->page, buf->page_offset, buf1_len, priv->dma_buf_sz); /* Data payload appended into SKB */ @@ -4345,7 +4424,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) if (buf2_len) { dma_sync_single_for_cpu(priv->device, buf->sec_addr, - buf2_len, DMA_FROM_DEVICE); + buf2_len, dma_dir); skb_add_rx_frag(skb, skb_shinfo(skb)->nr_frags, buf->sec_page, 0, buf2_len, priv->dma_buf_sz); @@ -4503,6 +4582,11 @@ static int stmmac_change_mtu(struct net_device *dev, int new_mtu) return -EBUSY; } + if (stmmac_xdp_is_enabled(priv) && new_mtu > ETH_DATA_LEN) { + netdev_dbg(priv->dev, "Jumbo frames not supported for XDP\n"); + return -EINVAL; + } + new_mtu = STMMAC_ALIGN(new_mtu); /* If condition true, FIFO is too small or MTU too large */ @@ -4564,6 +4648,7 @@ static int stmmac_set_features(struct net_device *netdev, stmmac_rx_ipc(priv, priv->hw); sph_en = (priv->hw->rx_csum > 0) && priv->sph; + for (chan = 0; chan < priv->plat->rx_queues_to_use; chan++) stmmac_enable_sph(priv, priv->ioaddr, sph_en, chan); @@ -5299,6 +5384,18 @@ static int stmmac_vlan_rx_kill_vid(struct net_device *ndev, __be16 proto, u16 vi return ret; } +static int stmmac_bpf(struct net_device *dev, struct netdev_bpf *bpf) +{ + struct stmmac_priv *priv = netdev_priv(dev); + + switch (bpf->command) { + case XDP_SETUP_PROG: + return stmmac_xdp_set_prog(priv, bpf->prog, bpf->extack); + default: + return -EOPNOTSUPP; + } +} + static const struct net_device_ops stmmac_netdev_ops = { .ndo_open = stmmac_open, .ndo_start_xmit = stmmac_xmit, @@ -5317,6 +5414,7 @@ static const struct net_device_ops stmmac_netdev_ops = { .ndo_set_mac_address = stmmac_set_mac_address, .ndo_vlan_rx_add_vid = stmmac_vlan_rx_add_vid, .ndo_vlan_rx_kill_vid = stmmac_vlan_rx_kill_vid, + .ndo_bpf = stmmac_bpf, }; static void stmmac_reset_subtask(struct stmmac_priv *priv) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.c new file mode 100644 index 000000000000..bf38d231860b --- /dev/null +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.c @@ -0,0 +1,40 @@ +// SPDX-License-Identifier: GPL-2.0 +/* Copyright (c) 2021, Intel Corporation. */ + +#include "stmmac.h" +#include "stmmac_xdp.h" + +int stmmac_xdp_set_prog(struct stmmac_priv *priv, struct bpf_prog *prog, + struct netlink_ext_ack *extack) +{ + struct net_device *dev = priv->dev; + struct bpf_prog *old_prog; + bool need_update; + bool if_running; + + if_running = netif_running(dev); + + if (prog && dev->mtu > ETH_DATA_LEN) { + /* For now, the driver doesn't support XDP functionality with + * jumbo frames so we return error. + */ + NL_SET_ERR_MSG_MOD(extack, "Jumbo frames not supported"); + return -EOPNOTSUPP; + } + + need_update = !!priv->xdp_prog != !!prog; + if (if_running && need_update) + stmmac_release(dev); + + old_prog = xchg(&priv->xdp_prog, prog); + if (old_prog) + bpf_prog_put(old_prog); + + /* Disable RX SPH for XDP operation */ + priv->sph = priv->sph_cap && !stmmac_xdp_is_enabled(priv); + + if (if_running && need_update) + stmmac_open(dev); + + return 0; +} diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.h b/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.h new file mode 100644 index 000000000000..93948569d92a --- /dev/null +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_xdp.h @@ -0,0 +1,12 @@ +/* SPDX-License-Identifier: GPL-2.0 */ +/* Copyright (c) 2021, Intel Corporation. */ + +#ifndef _STMMAC_XDP_H_ +#define _STMMAC_XDP_H_ + +#define STMMAC_MAX_RX_BUF_SIZE(num) (((num) * PAGE_SIZE) - XDP_PACKET_HEADROOM) + +int stmmac_xdp_set_prog(struct stmmac_priv *priv, struct bpf_prog *prog, + struct netlink_ext_ack *extack); + +#endif /* _STMMAC_XDP_H_ */ From patchwork Thu Apr 1 02:11:16 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176873 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id BDEA3C433B4 for ; Thu, 1 Apr 2021 02:09:03 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id 8DCCA6109F for ; Thu, 1 Apr 2021 02:09:03 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S233345AbhDACIc (ORCPT ); Wed, 31 Mar 2021 22:08:32 -0400 Received: from mga17.intel.com ([192.55.52.151]:22177 "EHLO mga17.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S233167AbhDACID (ORCPT ); Wed, 31 Mar 2021 22:08:03 -0400 IronPort-SDR: IkEPhQvYiu9IDjOxdi2MDT5c0o6hz6eic7A33P7WGkSmljUIuxu8DF6FFiUWYISDqrIHyXfEba /3wmotAv5KZQ== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="172163672" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="172163672" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by fmsmga107.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:08:02 -0700 IronPort-SDR: zcu4Yn+LhyRi83eGTXv95/rznlWErmfLk8y7c60WNPm8OfwiwzTBka8CHKReuLXcpCZ7R/p5WN 4y32HrWKfV6g== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004267" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:07:57 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 5/6] net: stmmac: Add support for XDP_TX action Date: Thu, 1 Apr 2021 10:11:16 +0800 Message-Id: <20210401021117.13360-6-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org This patch adds support for XDP_TX action which enables XDP program to transmit back received frames. This patch has been tested with the "xdp2" app located in samples/bpf dir. The DUT receives burst traffic packet generated using pktgen script 'pktgen_sample03_burst_single_flow.sh'. v4: Moved stmmac_tx_timer_arm() to be done once at the end of NAPI RX. Fixed stmmac_xdp_xmit_back() to return STMMAC_XDP_CONSUMED if XDP buffer to frame conversion fails. Thanks to Jakub's input. v3: Added 'nq->trans_start = jiffies' to avoid TX time-out as we are sharing TX queue between slow path and XDP. Thanks to Jakub Kicinski for pointing out. Signed-off-by: Ong Boon Leong --- drivers/net/ethernet/stmicro/stmmac/stmmac.h | 12 +- .../net/ethernet/stmicro/stmmac/stmmac_main.c | 238 ++++++++++++++++-- 2 files changed, 232 insertions(+), 18 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h index e72224c8fbac..a93e22a6be59 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h @@ -36,12 +36,18 @@ struct stmmac_resources { int tx_irq[MTL_MAX_TX_QUEUES]; }; +enum stmmac_txbuf_type { + STMMAC_TXBUF_T_SKB, + STMMAC_TXBUF_T_XDP_TX, +}; + struct stmmac_tx_info { dma_addr_t buf; bool map_as_page; unsigned len; bool last_segment; bool is_jumbo; + enum stmmac_txbuf_type buf_type; }; #define STMMAC_TBS_AVAIL BIT(0) @@ -57,7 +63,10 @@ struct stmmac_tx_queue { struct dma_extended_desc *dma_etx ____cacheline_aligned_in_smp; struct dma_edesc *dma_entx; struct dma_desc *dma_tx; - struct sk_buff **tx_skbuff; + union { + struct sk_buff **tx_skbuff; + struct xdp_frame **xdpf; + }; struct stmmac_tx_info *tx_skbuff_dma; unsigned int cur_tx; unsigned int dirty_tx; @@ -77,6 +86,7 @@ struct stmmac_rx_buffer { struct stmmac_rx_queue { u32 rx_count_frames; u32 queue_index; + struct xdp_rxq_info xdp_rxq; struct page_pool *page_pool; struct stmmac_rx_buffer *buf_pool; struct stmmac_priv *priv_data; diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 0dad8ab93eb5..65163b51f8ad 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -71,6 +71,7 @@ MODULE_PARM_DESC(phyaddr, "Physical device address"); #define STMMAC_XDP_PASS 0 #define STMMAC_XDP_CONSUMED BIT(0) +#define STMMAC_XDP_TX BIT(1) static int flow_ctrl = FLOW_AUTO; module_param(flow_ctrl, int, 0644); @@ -1442,7 +1443,8 @@ static void stmmac_free_tx_buffer(struct stmmac_priv *priv, u32 queue, int i) { struct stmmac_tx_queue *tx_q = &priv->tx_queue[queue]; - if (tx_q->tx_skbuff_dma[i].buf) { + if (tx_q->tx_skbuff_dma[i].buf && + tx_q->tx_skbuff_dma[i].buf_type != STMMAC_TXBUF_T_XDP_TX) { if (tx_q->tx_skbuff_dma[i].map_as_page) dma_unmap_page(priv->device, tx_q->tx_skbuff_dma[i].buf, @@ -1455,12 +1457,20 @@ static void stmmac_free_tx_buffer(struct stmmac_priv *priv, u32 queue, int i) DMA_TO_DEVICE); } - if (tx_q->tx_skbuff[i]) { + if (tx_q->xdpf[i] && + tx_q->tx_skbuff_dma[i].buf_type == STMMAC_TXBUF_T_XDP_TX) { + xdp_return_frame(tx_q->xdpf[i]); + tx_q->xdpf[i] = NULL; + } + + if (tx_q->tx_skbuff[i] && + tx_q->tx_skbuff_dma[i].buf_type == STMMAC_TXBUF_T_SKB) { dev_kfree_skb_any(tx_q->tx_skbuff[i]); tx_q->tx_skbuff[i] = NULL; - tx_q->tx_skbuff_dma[i].buf = 0; - tx_q->tx_skbuff_dma[i].map_as_page = false; } + + tx_q->tx_skbuff_dma[i].buf = 0; + tx_q->tx_skbuff_dma[i].map_as_page = false; } /** @@ -1568,6 +1578,7 @@ static int init_dma_rx_desc_rings(struct net_device *dev, gfp_t flags) for (queue = 0; queue < rx_count; queue++) { struct stmmac_rx_queue *rx_q = &priv->rx_queue[queue]; + int ret; netif_dbg(priv, probe, priv->dev, "(%s) dma_rx_phy=0x%08x\n", __func__, @@ -1575,6 +1586,14 @@ static int init_dma_rx_desc_rings(struct net_device *dev, gfp_t flags) stmmac_clear_rx_descriptors(priv, queue); + WARN_ON(xdp_rxq_info_reg_mem_model(&rx_q->xdp_rxq, + MEM_TYPE_PAGE_POOL, + rx_q->page_pool)); + + netdev_info(priv->dev, + "Register MEM_TYPE_PAGE_POOL RxQ-%d\n", + rx_q->queue_index); + for (i = 0; i < priv->dma_rx_size; i++) { struct dma_desc *p; @@ -1775,6 +1794,9 @@ static void free_dma_rx_desc_resources(struct stmmac_priv *priv) sizeof(struct dma_extended_desc), rx_q->dma_erx, rx_q->dma_rx_phy); + if (xdp_rxq_info_is_reg(&rx_q->xdp_rxq)) + xdp_rxq_info_unreg(&rx_q->xdp_rxq); + kfree(rx_q->buf_pool); if (rx_q->page_pool) page_pool_destroy(rx_q->page_pool); @@ -1837,8 +1859,10 @@ static int alloc_dma_rx_desc_resources(struct stmmac_priv *priv) /* RX queues buffers and DMA */ for (queue = 0; queue < rx_count; queue++) { struct stmmac_rx_queue *rx_q = &priv->rx_queue[queue]; + struct stmmac_channel *ch = &priv->channel[queue]; struct page_pool_params pp_params = { 0 }; unsigned int num_pages; + int ret; rx_q->queue_index = queue; rx_q->priv_data = priv; @@ -1884,6 +1908,14 @@ static int alloc_dma_rx_desc_resources(struct stmmac_priv *priv) if (!rx_q->dma_rx) goto err_dma; } + + ret = xdp_rxq_info_reg(&rx_q->xdp_rxq, priv->dev, + rx_q->queue_index, + ch->rx_napi.napi_id); + if (ret) { + netdev_err(priv->dev, "Failed to register xdp rxq info\n"); + goto err_dma; + } } return 0; @@ -1985,11 +2017,13 @@ static int alloc_dma_desc_resources(struct stmmac_priv *priv) */ static void free_dma_desc_resources(struct stmmac_priv *priv) { - /* Release the DMA RX socket buffers */ - free_dma_rx_desc_resources(priv); - /* Release the DMA TX socket buffers */ free_dma_tx_desc_resources(priv); + + /* Release the DMA RX socket buffers later + * to ensure all pending XDP_TX buffers are returned. + */ + free_dma_rx_desc_resources(priv); } /** @@ -2181,10 +2215,22 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue) entry = tx_q->dirty_tx; while ((entry != tx_q->cur_tx) && (count < budget)) { - struct sk_buff *skb = tx_q->tx_skbuff[entry]; + struct xdp_frame *xdpf; + struct sk_buff *skb; struct dma_desc *p; int status; + if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_TX) { + xdpf = tx_q->xdpf[entry]; + skb = NULL; + } else if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_SKB) { + xdpf = NULL; + skb = tx_q->tx_skbuff[entry]; + } else { + xdpf = NULL; + skb = NULL; + } + if (priv->extend_desc) p = (struct dma_desc *)(tx_q->dma_etx + entry); else if (tx_q->tbs & STMMAC_TBS_AVAIL) @@ -2214,10 +2260,12 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue) priv->dev->stats.tx_packets++; priv->xstats.tx_pkt_n++; } - stmmac_get_tx_hwtstamp(priv, p, skb); + if (skb) + stmmac_get_tx_hwtstamp(priv, p, skb); } - if (likely(tx_q->tx_skbuff_dma[entry].buf)) { + if (likely(tx_q->tx_skbuff_dma[entry].buf && + tx_q->tx_skbuff_dma[entry].buf_type != STMMAC_TXBUF_T_XDP_TX)) { if (tx_q->tx_skbuff_dma[entry].map_as_page) dma_unmap_page(priv->device, tx_q->tx_skbuff_dma[entry].buf, @@ -2238,11 +2286,19 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue) tx_q->tx_skbuff_dma[entry].last_segment = false; tx_q->tx_skbuff_dma[entry].is_jumbo = false; - if (likely(skb != NULL)) { - pkts_compl++; - bytes_compl += skb->len; - dev_consume_skb_any(skb); - tx_q->tx_skbuff[entry] = NULL; + if (xdpf && + tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_TX) { + xdp_return_frame_rx_napi(xdpf); + tx_q->xdpf[entry] = NULL; + } + + if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_SKB) { + if (likely(skb)) { + pkts_compl++; + bytes_compl += skb->len; + dev_consume_skb_any(skb); + tx_q->tx_skbuff[entry] = NULL; + } } stmmac_release_tx_desc(priv, p, priv->mode); @@ -3667,6 +3723,8 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev) tx_q->tx_skbuff_dma[first_entry].buf = des; tx_q->tx_skbuff_dma[first_entry].len = skb_headlen(skb); + tx_q->tx_skbuff_dma[first_entry].map_as_page = false; + tx_q->tx_skbuff_dma[first_entry].buf_type = STMMAC_TXBUF_T_SKB; if (priv->dma_cap.addr64 <= 32) { first->des0 = cpu_to_le32(des); @@ -3702,12 +3760,14 @@ static netdev_tx_t stmmac_tso_xmit(struct sk_buff *skb, struct net_device *dev) tx_q->tx_skbuff_dma[tx_q->cur_tx].buf = des; tx_q->tx_skbuff_dma[tx_q->cur_tx].len = skb_frag_size(frag); tx_q->tx_skbuff_dma[tx_q->cur_tx].map_as_page = true; + tx_q->tx_skbuff_dma[tx_q->cur_tx].buf_type = STMMAC_TXBUF_T_SKB; } tx_q->tx_skbuff_dma[tx_q->cur_tx].last_segment = true; /* Only the last descriptor gets to point to the skb. */ tx_q->tx_skbuff[tx_q->cur_tx] = skb; + tx_q->tx_skbuff_dma[tx_q->cur_tx].buf_type = STMMAC_TXBUF_T_SKB; /* Manage tx mitigation */ tx_packets = (tx_q->cur_tx + 1) - first_tx; @@ -3914,6 +3974,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev) tx_q->tx_skbuff_dma[entry].map_as_page = true; tx_q->tx_skbuff_dma[entry].len = len; tx_q->tx_skbuff_dma[entry].last_segment = last_segment; + tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_SKB; /* Prepare the descriptor and set the own bit too */ stmmac_prepare_tx_desc(priv, desc, 0, len, csum_insertion, @@ -3922,6 +3983,7 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev) /* Only the last descriptor gets to point to the skb. */ tx_q->tx_skbuff[entry] = skb; + tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_SKB; /* According to the coalesce parameter the IC bit for the latest * segment is reset and the timer re-started to clean the tx status. @@ -4000,6 +4062,8 @@ static netdev_tx_t stmmac_xmit(struct sk_buff *skb, struct net_device *dev) goto dma_map_err; tx_q->tx_skbuff_dma[first_entry].buf = des; + tx_q->tx_skbuff_dma[first_entry].buf_type = STMMAC_TXBUF_T_SKB; + tx_q->tx_skbuff_dma[first_entry].map_as_page = false; stmmac_set_desc_addr(priv, first, des); @@ -4181,6 +4245,110 @@ static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, return plen - len; } +static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue, + struct xdp_frame *xdpf) +{ + struct stmmac_tx_queue *tx_q = &priv->tx_queue[queue]; + struct page *page = virt_to_page(xdpf->data); + unsigned int entry = tx_q->cur_tx; + struct dma_desc *tx_desc; + dma_addr_t dma_addr; + bool set_ic; + + if (stmmac_tx_avail(priv, queue) < STMMAC_TX_THRESH(priv)) + return STMMAC_XDP_CONSUMED; + + if (likely(priv->extend_desc)) + tx_desc = (struct dma_desc *)(tx_q->dma_etx + entry); + else if (tx_q->tbs & STMMAC_TBS_AVAIL) + tx_desc = &tx_q->dma_entx[entry].basic; + else + tx_desc = tx_q->dma_tx + entry; + + dma_addr = page_pool_get_dma_addr(page) + sizeof(*xdpf) + + xdpf->headroom; + dma_sync_single_for_device(priv->device, dma_addr, + xdpf->len, DMA_BIDIRECTIONAL); + + tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_XDP_TX; + + tx_q->tx_skbuff_dma[entry].buf = dma_addr; + tx_q->tx_skbuff_dma[entry].map_as_page = false; + tx_q->tx_skbuff_dma[entry].len = xdpf->len; + tx_q->tx_skbuff_dma[entry].last_segment = true; + tx_q->tx_skbuff_dma[entry].is_jumbo = false; + + tx_q->xdpf[entry] = xdpf; + + stmmac_set_desc_addr(priv, tx_desc, dma_addr); + + stmmac_prepare_tx_desc(priv, tx_desc, 1, xdpf->len, + true, priv->mode, true, true, + xdpf->len); + + tx_q->tx_count_frames++; + + if (tx_q->tx_count_frames % priv->tx_coal_frames[queue] == 0) + set_ic = true; + else + set_ic = false; + + if (set_ic) { + tx_q->tx_count_frames = 0; + stmmac_set_tx_ic(priv, tx_desc); + priv->xstats.tx_set_ic_bit++; + } + + stmmac_enable_dma_transmission(priv, priv->ioaddr); + + entry = STMMAC_GET_ENTRY(entry, priv->dma_tx_size); + tx_q->cur_tx = entry; + + return STMMAC_XDP_TX; +} + +static int stmmac_xdp_get_tx_queue(struct stmmac_priv *priv, + int cpu) +{ + int index = cpu; + + if (unlikely(index < 0)) + index = 0; + + while (index >= priv->plat->tx_queues_to_use) + index -= priv->plat->tx_queues_to_use; + + return index; +} + +static int stmmac_xdp_xmit_back(struct stmmac_priv *priv, + struct xdp_buff *xdp) +{ + struct xdp_frame *xdpf = xdp_convert_buff_to_frame(xdp); + int cpu = smp_processor_id(); + struct netdev_queue *nq; + int queue; + int res; + + if (unlikely(!xdpf)) + return STMMAC_XDP_CONSUMED; + + queue = stmmac_xdp_get_tx_queue(priv, cpu); + nq = netdev_get_tx_queue(priv->dev, queue); + + __netif_tx_lock(nq, cpu); + /* Avoids TX time-out as we are sharing with slow path */ + nq->trans_start = jiffies; + + res = stmmac_xdp_xmit_xdpf(priv, queue, xdpf); + if (res == STMMAC_XDP_TX) + stmmac_flush_tx_descriptors(priv, queue); + + __netif_tx_unlock(nq); + + return res; +} + static struct sk_buff *stmmac_xdp_run_prog(struct stmmac_priv *priv, struct xdp_buff *xdp) { @@ -4201,6 +4369,9 @@ static struct sk_buff *stmmac_xdp_run_prog(struct stmmac_priv *priv, case XDP_PASS: res = STMMAC_XDP_PASS; break; + case XDP_TX: + res = stmmac_xdp_xmit_back(priv, xdp); + break; default: bpf_warn_invalid_xdp_action(act); fallthrough; @@ -4217,6 +4388,18 @@ static struct sk_buff *stmmac_xdp_run_prog(struct stmmac_priv *priv, return ERR_PTR(-res); } +static void stmmac_finalize_xdp_rx(struct stmmac_priv *priv, + int xdp_status) +{ + int cpu = smp_processor_id(); + int queue; + + queue = stmmac_xdp_get_tx_queue(priv, cpu); + + if (xdp_status & STMMAC_XDP_TX) + stmmac_tx_timer_arm(priv, queue); +} + /** * stmmac_rx - manage the receive process * @priv: driver private structure @@ -4236,6 +4419,7 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) unsigned int desc_size; struct sk_buff *skb = NULL; struct xdp_buff xdp; + int xdp_status = 0; int buf_sz; dma_dir = page_pool_get_dma_dir(rx_q->page_pool); @@ -4357,6 +4541,8 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) } if (!skb) { + unsigned int pre_len, sync_len; + dma_sync_single_for_cpu(priv->device, buf->addr, buf1_len, dma_dir); @@ -4365,16 +4551,26 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) xdp.data_hard_start = page_address(buf->page); xdp_set_data_meta_invalid(&xdp); xdp.frame_sz = buf_sz; + xdp.rxq = &rx_q->xdp_rxq; + pre_len = xdp.data_end - xdp.data_hard_start - + buf->page_offset; skb = stmmac_xdp_run_prog(priv, &xdp); + /* Due xdp_adjust_tail: DMA sync for_device + * cover max len CPU touch + */ + sync_len = xdp.data_end - xdp.data_hard_start - + buf->page_offset; + sync_len = max(sync_len, pre_len); /* For Not XDP_PASS verdict */ if (IS_ERR(skb)) { unsigned int xdp_res = -PTR_ERR(skb); if (xdp_res & STMMAC_XDP_CONSUMED) { - page_pool_recycle_direct(rx_q->page_pool, - buf->page); + page_pool_put_page(rx_q->page_pool, + virt_to_head_page(xdp.data), + sync_len, true); buf->page = NULL; priv->dev->stats.rx_dropped++; @@ -4388,6 +4584,12 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) count++; continue; + } else if (xdp_res & STMMAC_XDP_TX) { + xdp_status |= xdp_res; + buf->page = NULL; + skb = NULL; + count++; + continue; } } } @@ -4470,6 +4672,8 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) rx_q->state.len = len; } + stmmac_finalize_xdp_rx(priv, xdp_status); + stmmac_rx_refill(priv, queue); priv->xstats.rx_pkt_n += count; From patchwork Thu Apr 1 02:11:17 2021 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: Ong Boon Leong X-Patchwork-Id: 12176875 X-Patchwork-Delegate: kuba@kernel.org Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org X-Spam-Level: X-Spam-Status: No, score=-16.8 required=3.0 tests=BAYES_00, HEADER_FROM_DIFFERENT_DOMAINS,INCLUDES_CR_TRAILER,INCLUDES_PATCH, MAILING_LIST_MULTI,SPF_HELO_NONE,SPF_PASS,URIBL_BLOCKED,USER_AGENT_GIT autolearn=ham autolearn_force=no version=3.4.0 Received: from mail.kernel.org (mail.kernel.org [198.145.29.99]) by smtp.lore.kernel.org (Postfix) with ESMTP id 06C50C43461 for ; Thu, 1 Apr 2021 02:09:05 +0000 (UTC) Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by mail.kernel.org (Postfix) with ESMTP id C21856109F for ; Thu, 1 Apr 2021 02:09:04 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S233361AbhDACId (ORCPT ); Wed, 31 Mar 2021 22:08:33 -0400 Received: from mga17.intel.com ([192.55.52.151]:22177 "EHLO mga17.intel.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S233305AbhDACII (ORCPT ); Wed, 31 Mar 2021 22:08:08 -0400 IronPort-SDR: 49dRrPkplTcfkmu5zeC9DjlJ8W0K1h9YxcYmD9TLPY4VxxL+M6dwWK827F7s7EFlnbeW8qefjq odqaRwrwNVvA== X-IronPort-AV: E=McAfee;i="6000,8403,9940"; a="172163702" X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="172163702" Received: from orsmga004.jf.intel.com ([10.7.209.38]) by fmsmga107.fm.intel.com with ESMTP/TLS/ECDHE-RSA-AES256-GCM-SHA384; 31 Mar 2021 19:08:08 -0700 IronPort-SDR: 88+4Hve+PQjv7Zhmtn6b1OBdEapIKl9MYDgo4SP1QnZip5D+7+aIjWSwO5oGA78EJGiB1A/7K7 NF30+/i37/tA== X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.81,295,1610438400"; d="scan'208";a="528004281" Received: from glass.png.intel.com ([10.158.65.59]) by orsmga004.jf.intel.com with ESMTP; 31 Mar 2021 19:08:02 -0700 From: Ong Boon Leong To: Giuseppe Cavallaro , Alexandre Torgue , Jose Abreu , "David S . Miller" , Jakub Kicinski , Alexei Starovoitov , Daniel Borkmann , Jesper Dangaard Brouer , John Fastabend Cc: Maxime Coquelin , Andrii Nakryiko , Martin KaFai Lau , Song Liu , Yonghong Song , KP Singh , netdev@vger.kernel.org, linux-stm32@st-md-mailman.stormreply.com, linux-arm-kernel@lists.infradead.org, linux-kernel@vger.kernel.org, bpf@vger.kernel.org, Ong Boon Leong Subject: [PATCH net-next v4 6/6] net: stmmac: Add support for XDP_REDIRECT action Date: Thu, 1 Apr 2021 10:11:17 +0800 Message-Id: <20210401021117.13360-7-boon.leong.ong@intel.com> X-Mailer: git-send-email 2.25.1 In-Reply-To: <20210401021117.13360-1-boon.leong.ong@intel.com> References: <20210401021117.13360-1-boon.leong.ong@intel.com> MIME-Version: 1.0 Precedence: bulk List-ID: X-Mailing-List: bpf@vger.kernel.org X-Patchwork-Delegate: kuba@kernel.org This patch adds the support of XDP_REDIRECT to another remote cpu for further action. It also implements ndo_xdp_xmit ops, enabling the driver to transmit packets forwarded to it by XDP program running on another interface. This patch has been tested using "xdp_redirect_cpu" for XDP_REDIRECT + drop testing. It also been tested with "xdp_redirect" sample app which can be used to exercise ndo_xdp_xmit ops. The burst traffics are generated using pktgen_sample03_burst_single_flow.sh in samples/pktgen directory. v4: Move xdp_do_flush() processing into stmmac_finalize_xdp_rx() and combined the XDP verdict of XDP TX and REDIRECT together. v3: Added 'nq->trans_start = jiffies' to avoid TX time-out as we are sharing TX queue between slow path and XDP. Thanks to Jakub Kicinski for point out. Signed-off-by: Ong Boon Leong --- drivers/net/ethernet/stmicro/stmmac/stmmac.h | 1 + .../net/ethernet/stmicro/stmmac/stmmac_main.c | 94 ++++++++++++++++--- 2 files changed, 84 insertions(+), 11 deletions(-) diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac.h b/drivers/net/ethernet/stmicro/stmmac/stmmac.h index a93e22a6be59..c49debb62b05 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac.h +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac.h @@ -39,6 +39,7 @@ struct stmmac_resources { enum stmmac_txbuf_type { STMMAC_TXBUF_T_SKB, STMMAC_TXBUF_T_XDP_TX, + STMMAC_TXBUF_T_XDP_NDO, }; struct stmmac_tx_info { diff --git a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c index 65163b51f8ad..77285646c5fc 100644 --- a/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c +++ b/drivers/net/ethernet/stmicro/stmmac/stmmac_main.c @@ -72,6 +72,7 @@ MODULE_PARM_DESC(phyaddr, "Physical device address"); #define STMMAC_XDP_PASS 0 #define STMMAC_XDP_CONSUMED BIT(0) #define STMMAC_XDP_TX BIT(1) +#define STMMAC_XDP_REDIRECT BIT(2) static int flow_ctrl = FLOW_AUTO; module_param(flow_ctrl, int, 0644); @@ -1458,7 +1459,8 @@ static void stmmac_free_tx_buffer(struct stmmac_priv *priv, u32 queue, int i) } if (tx_q->xdpf[i] && - tx_q->tx_skbuff_dma[i].buf_type == STMMAC_TXBUF_T_XDP_TX) { + (tx_q->tx_skbuff_dma[i].buf_type == STMMAC_TXBUF_T_XDP_TX || + tx_q->tx_skbuff_dma[i].buf_type == STMMAC_TXBUF_T_XDP_NDO)) { xdp_return_frame(tx_q->xdpf[i]); tx_q->xdpf[i] = NULL; } @@ -2220,7 +2222,8 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue) struct dma_desc *p; int status; - if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_TX) { + if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_TX || + tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_NDO) { xdpf = tx_q->xdpf[entry]; skb = NULL; } else if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_SKB) { @@ -2292,6 +2295,12 @@ static int stmmac_tx_clean(struct stmmac_priv *priv, int budget, u32 queue) tx_q->xdpf[entry] = NULL; } + if (xdpf && + tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_XDP_NDO) { + xdp_return_frame(xdpf); + tx_q->xdpf[entry] = NULL; + } + if (tx_q->tx_skbuff_dma[entry].buf_type == STMMAC_TXBUF_T_SKB) { if (likely(skb)) { pkts_compl++; @@ -4246,10 +4255,9 @@ static unsigned int stmmac_rx_buf2_len(struct stmmac_priv *priv, } static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue, - struct xdp_frame *xdpf) + struct xdp_frame *xdpf, bool dma_map) { struct stmmac_tx_queue *tx_q = &priv->tx_queue[queue]; - struct page *page = virt_to_page(xdpf->data); unsigned int entry = tx_q->cur_tx; struct dma_desc *tx_desc; dma_addr_t dma_addr; @@ -4265,12 +4273,23 @@ static int stmmac_xdp_xmit_xdpf(struct stmmac_priv *priv, int queue, else tx_desc = tx_q->dma_tx + entry; - dma_addr = page_pool_get_dma_addr(page) + sizeof(*xdpf) + - xdpf->headroom; - dma_sync_single_for_device(priv->device, dma_addr, - xdpf->len, DMA_BIDIRECTIONAL); + if (dma_map) { + dma_addr = dma_map_single(priv->device, xdpf->data, + xdpf->len, DMA_TO_DEVICE); + if (dma_mapping_error(priv->device, dma_addr)) + return STMMAC_XDP_CONSUMED; + + tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_XDP_NDO; + } else { + struct page *page = virt_to_page(xdpf->data); + + dma_addr = page_pool_get_dma_addr(page) + sizeof(*xdpf) + + xdpf->headroom; + dma_sync_single_for_device(priv->device, dma_addr, + xdpf->len, DMA_BIDIRECTIONAL); - tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_XDP_TX; + tx_q->tx_skbuff_dma[entry].buf_type = STMMAC_TXBUF_T_XDP_TX; + } tx_q->tx_skbuff_dma[entry].buf = dma_addr; tx_q->tx_skbuff_dma[entry].map_as_page = false; @@ -4340,7 +4359,7 @@ static int stmmac_xdp_xmit_back(struct stmmac_priv *priv, /* Avoids TX time-out as we are sharing with slow path */ nq->trans_start = jiffies; - res = stmmac_xdp_xmit_xdpf(priv, queue, xdpf); + res = stmmac_xdp_xmit_xdpf(priv, queue, xdpf, false); if (res == STMMAC_XDP_TX) stmmac_flush_tx_descriptors(priv, queue); @@ -4372,6 +4391,12 @@ static struct sk_buff *stmmac_xdp_run_prog(struct stmmac_priv *priv, case XDP_TX: res = stmmac_xdp_xmit_back(priv, xdp); break; + case XDP_REDIRECT: + if (xdp_do_redirect(priv->dev, xdp, prog) < 0) + res = STMMAC_XDP_CONSUMED; + else + res = STMMAC_XDP_REDIRECT; + break; default: bpf_warn_invalid_xdp_action(act); fallthrough; @@ -4398,6 +4423,9 @@ static void stmmac_finalize_xdp_rx(struct stmmac_priv *priv, if (xdp_status & STMMAC_XDP_TX) stmmac_tx_timer_arm(priv, queue); + + if (xdp_status & STMMAC_XDP_REDIRECT) + xdp_do_flush(); } /** @@ -4584,7 +4612,8 @@ static int stmmac_rx(struct stmmac_priv *priv, int limit, u32 queue) count++; continue; - } else if (xdp_res & STMMAC_XDP_TX) { + } else if (xdp_res & (STMMAC_XDP_TX | + STMMAC_XDP_REDIRECT)) { xdp_status |= xdp_res; buf->page = NULL; skb = NULL; @@ -5600,6 +5629,48 @@ static int stmmac_bpf(struct net_device *dev, struct netdev_bpf *bpf) } } +static int stmmac_xdp_xmit(struct net_device *dev, int num_frames, + struct xdp_frame **frames, u32 flags) +{ + struct stmmac_priv *priv = netdev_priv(dev); + int cpu = smp_processor_id(); + struct netdev_queue *nq; + int i, nxmit = 0; + int queue; + + if (unlikely(test_bit(STMMAC_DOWN, &priv->state))) + return -ENETDOWN; + + if (unlikely(flags & ~XDP_XMIT_FLAGS_MASK)) + return -EINVAL; + + queue = stmmac_xdp_get_tx_queue(priv, cpu); + nq = netdev_get_tx_queue(priv->dev, queue); + + __netif_tx_lock(nq, cpu); + /* Avoids TX time-out as we are sharing with slow path */ + nq->trans_start = jiffies; + + for (i = 0; i < num_frames; i++) { + int res; + + res = stmmac_xdp_xmit_xdpf(priv, queue, frames[i], true); + if (res == STMMAC_XDP_CONSUMED) + break; + + nxmit++; + } + + if (flags & XDP_XMIT_FLUSH) { + stmmac_flush_tx_descriptors(priv, queue); + stmmac_tx_timer_arm(priv, queue); + } + + __netif_tx_unlock(nq); + + return nxmit; +} + static const struct net_device_ops stmmac_netdev_ops = { .ndo_open = stmmac_open, .ndo_start_xmit = stmmac_xmit, @@ -5619,6 +5690,7 @@ static const struct net_device_ops stmmac_netdev_ops = { .ndo_vlan_rx_add_vid = stmmac_vlan_rx_add_vid, .ndo_vlan_rx_kill_vid = stmmac_vlan_rx_kill_vid, .ndo_bpf = stmmac_bpf, + .ndo_xdp_xmit = stmmac_xdp_xmit, }; static void stmmac_reset_subtask(struct stmmac_priv *priv)