From patchwork Fri Sep 9 21:29:38 2016 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: John Fastabend X-Patchwork-Id: 668276 X-Patchwork-Delegate: jeffrey.t.kirsher@intel.com Return-Path: X-Original-To: incoming@patchwork.ozlabs.org Delivered-To: patchwork-incoming@bilbo.ozlabs.org Received: from fraxinus.osuosl.org (smtp4.osuosl.org [140.211.166.137]) (using TLSv1.2 with cipher AECDH-AES256-SHA (256/256 bits)) (No client certificate requested) by ozlabs.org (Postfix) with ESMTPS id 3sW9Mr02y3z9s1h for ; Sat, 10 Sep 2016 07:29:59 +1000 (AEST) Authentication-Results: ozlabs.org; dkim=fail reason="signature verification failed" (2048-bit key; unprotected) header.d=gmail.com header.i=@gmail.com header.b=1Gd9YMhY; dkim-atps=neutral Received: from localhost (localhost [127.0.0.1]) by fraxinus.osuosl.org (Postfix) with ESMTP id 53A8B896A7; Fri, 9 Sep 2016 21:29:58 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from fraxinus.osuosl.org ([127.0.0.1]) by localhost (.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id Kkp3F5pmA7k0; Fri, 9 Sep 2016 21:29:56 +0000 (UTC) Received: from ash.osuosl.org (ash.osuosl.org [140.211.166.34]) by fraxinus.osuosl.org (Postfix) with ESMTP id 06B828931E; Fri, 9 Sep 2016 21:29:56 +0000 (UTC) X-Original-To: intel-wired-lan@lists.osuosl.org Delivered-To: intel-wired-lan@lists.osuosl.org Received: from silver.osuosl.org (smtp3.osuosl.org [140.211.166.136]) by ash.osuosl.org (Postfix) with ESMTP id AF9781C2B2B for ; Fri, 9 Sep 2016 21:29:55 +0000 (UTC) Received: from localhost (localhost [127.0.0.1]) by silver.osuosl.org (Postfix) with ESMTP id A7AA33146C for ; Fri, 9 Sep 2016 21:29:55 +0000 (UTC) X-Virus-Scanned: amavisd-new at osuosl.org Received: from silver.osuosl.org ([127.0.0.1]) by localhost (.osuosl.org [127.0.0.1]) (amavisd-new, port 10024) with ESMTP id 4IEckypbVYq7 for ; Fri, 9 Sep 2016 21:29:55 +0000 (UTC) X-Greylist: domain auto-whitelisted by SQLgrey-1.7.6 Received: from mail-pa0-f66.google.com (mail-pa0-f66.google.com [209.85.220.66]) by silver.osuosl.org (Postfix) with ESMTPS id E3CE22E6CB for ; Fri, 9 Sep 2016 21:29:54 +0000 (UTC) Received: by mail-pa0-f66.google.com with SMTP id h5so4331304pao.0 for ; Fri, 09 Sep 2016 14:29:54 -0700 (PDT) DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=gmail.com; s=20120113; h=from:subject:to:cc:date:message-id:in-reply-to:references :user-agent:mime-version:content-transfer-encoding; bh=TNMgx1MJxPsUzMLx61d6dK/Njlq0CTY5QzcBr4F/zeA=; b=1Gd9YMhYsCgWzvAEnI3b5kygLAlChmV5GlpxpjgUP3emLSrVL91orS9rtDAGiZYwY9 iTBM5rrwGpnd2GjCH8TC/MmofcQGlHHze1EhSr8YrsjeUN1Uey9wQ9Ku9PPKMxtNG5+O dreMjw/41Q7oysiObAPZMUcXug4ybRQRs+jdDZSyMSm7n4lHF9tt23oVA/Qf0OnZwtkD 1OurShKnqOQLcE6HtydA0w6w0g4phQ2HHhXZHSfd16aH/nc2EI4GqMGoomP69wdYVscV ZZWbtzEq/cX+UesChyyCwjG9byCnQXm/jPcG1P3hQzdvu8SxvC47qmtw32CRKhqhO2uj PMTA== X-Google-DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=1e100.net; s=20130820; h=x-gm-message-state:from:subject:to:cc:date:message-id:in-reply-to :references:user-agent:mime-version:content-transfer-encoding; bh=TNMgx1MJxPsUzMLx61d6dK/Njlq0CTY5QzcBr4F/zeA=; b=gsqAm64+3m3qVxZLIHetafdjOy4tLIoU+QFb2hUlZuZDKGFXxALMeEqjMZQ6bzI2Vv 9wWRMox5Uo3je97kcA7MnOJQDFs1vbcTf2r9PrM8rRPfFPIEFCuiWbWRTO2tHC8caw4h nvMoCT+Le4DDvDyaj2tnivDUWo0CvxR+13IDvux9n63aIEJsSUZQepuCaHWXJp5a9ROZ 6+qR0DdeNaEN1fIsXCSxMU27cp165Bw7Gzt71MCvVEoKMd71rKS4GslqCmjl8t0UVO84 CTv41cDFNTZZOp6m121duybeAvHuUAEG2N/NWbg1uBx5b4jk+Py4wLWLzHHDWEYMufVd ehRw== X-Gm-Message-State: AE9vXwNUzILt0lZhsvJn3lYDiu913muWMUKKaEpqvaGD5GoXTYIfu9k2fhuUuF8TTuMr7w== X-Received: by 10.66.228.2 with SMTP id se2mr10097165pac.93.1473456594554; Fri, 09 Sep 2016 14:29:54 -0700 (PDT) Received: from [127.0.1.1] ([72.168.145.240]) by smtp.gmail.com with ESMTPSA id e72sm7384679pfb.49.2016.09.09.14.29.44 (version=TLS1_2 cipher=ECDHE-RSA-AES128-GCM-SHA256 bits=128/128); Fri, 09 Sep 2016 14:29:54 -0700 (PDT) From: John Fastabend X-Google-Original-From: John Fastabend To: bblanco@plumgrid.com, john.fastabend@gmail.com, alexei.starovoitov@gmail.com, jeffrey.t.kirsher@intel.com, brouer@redhat.com, davem@davemloft.net Date: Fri, 09 Sep 2016 14:29:38 -0700 Message-ID: <20160909212938.4001.40540.stgit@john-Precision-Tower-5810> In-Reply-To: <20160909212915.4001.25504.stgit@john-Precision-Tower-5810> References: <20160909212915.4001.25504.stgit@john-Precision-Tower-5810> User-Agent: StGit/0.17.1-dirty MIME-Version: 1.0 Cc: xiyou.wangcong@gmail.com, intel-wired-lan@lists.osuosl.org, u9012063@gmail.com, netdev@vger.kernel.org Subject: [Intel-wired-lan] [net-next PATCH v2 2/2] e1000: bundle xdp xmit routines X-BeenThere: intel-wired-lan@lists.osuosl.org X-Mailman-Version: 2.1.18-1 Precedence: list List-Id: Intel Wired Ethernet Linux Kernel Driver Development List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Errors-To: intel-wired-lan-bounces@lists.osuosl.org Sender: "Intel-wired-lan" e1000 supports a single TX queue so it is being shared with the stack when XDP runs XDP_TX action. This requires taking the xmit lock to ensure we don't corrupt the tx ring. To avoid taking and dropping the lock per packet this patch adds a bundling implementation to submit a bundle of packets to the xmit routine. I tested this patch running e1000 in a VM using KVM over a tap device using pktgen to generate traffic along with 'ping -f -l 100'. Suggested-by: Jesper Dangaard Brouer Signed-off-by: John Fastabend --- drivers/net/ethernet/intel/e1000/e1000.h | 10 +++ drivers/net/ethernet/intel/e1000/e1000_main.c | 81 +++++++++++++++++++------ 2 files changed, 71 insertions(+), 20 deletions(-) diff --git a/drivers/net/ethernet/intel/e1000/e1000.h b/drivers/net/ethernet/intel/e1000/e1000.h index 5cf8a0a..877b377 100644 --- a/drivers/net/ethernet/intel/e1000/e1000.h +++ b/drivers/net/ethernet/intel/e1000/e1000.h @@ -133,6 +133,8 @@ struct e1000_adapter; #define E1000_TX_QUEUE_WAKE 16 /* How many Rx Buffers do we bundle into one write to the hardware ? */ #define E1000_RX_BUFFER_WRITE 16 /* Must be power of 2 */ +/* How many XDP XMIT buffers to bundle into one xmit transaction */ +#define E1000_XDP_XMIT_BUNDLE_MAX E1000_RX_BUFFER_WRITE #define AUTO_ALL_MODES 0 #define E1000_EEPROM_82544_APM 0x0004 @@ -168,6 +170,11 @@ struct e1000_rx_buffer { dma_addr_t dma; }; +struct e1000_rx_buffer_bundle { + struct e1000_rx_buffer *buffer; + u32 length; +}; + struct e1000_tx_ring { /* pointer to the descriptor ring memory */ void *desc; @@ -206,6 +213,9 @@ struct e1000_rx_ring { struct e1000_rx_buffer *buffer_info; struct sk_buff *rx_skb_top; + /* array of XDP buffer information structs */ + struct e1000_rx_buffer_bundle *xdp_buffer; + /* cpu for rx queue */ int cpu; diff --git a/drivers/net/ethernet/intel/e1000/e1000_main.c b/drivers/net/ethernet/intel/e1000/e1000_main.c index 91d5c87..b985271 100644 --- a/drivers/net/ethernet/intel/e1000/e1000_main.c +++ b/drivers/net/ethernet/intel/e1000/e1000_main.c @@ -1738,10 +1738,18 @@ static int e1000_setup_rx_resources(struct e1000_adapter *adapter, struct pci_dev *pdev = adapter->pdev; int size, desc_len; + size = sizeof(struct e1000_rx_buffer_bundle) * + E1000_XDP_XMIT_BUNDLE_MAX; + rxdr->xdp_buffer = vzalloc(size); + if (!rxdr->xdp_buffer) + return -ENOMEM; + size = sizeof(struct e1000_rx_buffer) * rxdr->count; rxdr->buffer_info = vzalloc(size); - if (!rxdr->buffer_info) + if (!rxdr->buffer_info) { + vfree(rxdr->xdp_buffer); return -ENOMEM; + } desc_len = sizeof(struct e1000_rx_desc); @@ -1754,6 +1762,7 @@ static int e1000_setup_rx_resources(struct e1000_adapter *adapter, GFP_KERNEL); if (!rxdr->desc) { setup_rx_desc_die: + vfree(rxdr->xdp_buffer); vfree(rxdr->buffer_info); return -ENOMEM; } @@ -2087,6 +2096,9 @@ static void e1000_free_rx_resources(struct e1000_adapter *adapter, e1000_clean_rx_ring(adapter, rx_ring); + vfree(rx_ring->xdp_buffer); + rx_ring->xdp_buffer = NULL; + vfree(rx_ring->buffer_info); rx_ring->buffer_info = NULL; @@ -3369,33 +3381,52 @@ static void e1000_tx_map_rxpage(struct e1000_tx_ring *tx_ring, } static void e1000_xmit_raw_frame(struct e1000_rx_buffer *rx_buffer_info, - unsigned int len, - struct net_device *netdev, - struct e1000_adapter *adapter) + __u32 len, + struct e1000_adapter *adapter, + struct e1000_tx_ring *tx_ring) { - struct netdev_queue *txq = netdev_get_tx_queue(netdev, 0); - struct e1000_hw *hw = &adapter->hw; - struct e1000_tx_ring *tx_ring; - if (len > E1000_MAX_DATA_PER_TXD) return; + if (E1000_DESC_UNUSED(tx_ring) < 2) + return; + + e1000_tx_map_rxpage(tx_ring, rx_buffer_info, len); + e1000_tx_queue(adapter, tx_ring, 0/*tx_flags*/, 1); +} + +static void e1000_xdp_xmit_bundle(struct e1000_rx_buffer_bundle *buffer_info, + struct net_device *netdev, + struct e1000_adapter *adapter) +{ + struct netdev_queue *txq = netdev_get_tx_queue(netdev, 0); + struct e1000_tx_ring *tx_ring = adapter->tx_ring; + struct e1000_hw *hw = &adapter->hw; + int i = 0; + /* e1000 only support a single txq at the moment so the queue is being * shared with stack. To support this requires locking to ensure the * stack and XDP are not running at the same time. Devices with * multiple queues should allocate a separate queue space. + * + * To amortize the locking cost e1000 bundles the xmits and sends as + * many as possible until either running out of descriptors or failing. */ HARD_TX_LOCK(netdev, txq, smp_processor_id()); - tx_ring = adapter->tx_ring; - - if (E1000_DESC_UNUSED(tx_ring) < 2) { - HARD_TX_UNLOCK(netdev, txq); - return; + for (; i < E1000_XDP_XMIT_BUNDLE_MAX && buffer_info[i].buffer; i++) { + e1000_xmit_raw_frame(buffer_info[i].buffer, + buffer_info[i].length, + adapter, tx_ring); + buffer_info[i].buffer->rxbuf.page = NULL; + buffer_info[i].buffer = NULL; + buffer_info[i].length = 0; + i++; } - e1000_tx_map_rxpage(tx_ring, rx_buffer_info, len); - e1000_tx_queue(adapter, tx_ring, 0/*tx_flags*/, 1); + /* kick hardware to send bundle and return control back to the stack */ + writel(tx_ring->next_to_use, hw->hw_addr + tx_ring->tdt); + mmiowb(); writel(tx_ring->next_to_use, hw->hw_addr + tx_ring->tdt); mmiowb(); @@ -4281,9 +4312,10 @@ static bool e1000_clean_jumbo_rx_irq(struct e1000_adapter *adapter, struct bpf_prog *prog; u32 length; unsigned int i; - int cleaned_count = 0; + int cleaned_count = 0, xdp_xmit = 0; bool cleaned = false; unsigned int total_rx_bytes = 0, total_rx_packets = 0; + struct e1000_rx_buffer_bundle *xdp_bundle = rx_ring->xdp_buffer; rcu_read_lock(); /* rcu lock needed here to protect xdp programs */ prog = READ_ONCE(adapter->prog); @@ -4338,13 +4370,13 @@ static bool e1000_clean_jumbo_rx_irq(struct e1000_adapter *adapter, case XDP_PASS: break; case XDP_TX: + xdp_bundle[xdp_xmit].buffer = buffer_info; + xdp_bundle[xdp_xmit].length = length; dma_sync_single_for_device(&pdev->dev, dma, length, DMA_TO_DEVICE); - e1000_xmit_raw_frame(buffer_info, length, - netdev, adapter); - buffer_info->rxbuf.page = NULL; + xdp_xmit++; case XDP_DROP: default: /* re-use mapped page. keep buffer_info->dma @@ -4486,8 +4518,14 @@ next_desc: /* return some buffers to hardware, one at a time is too slow */ if (unlikely(cleaned_count >= E1000_RX_BUFFER_WRITE)) { + if (xdp_xmit) + e1000_xdp_xmit_bundle(xdp_bundle, + netdev, + adapter); + adapter->alloc_rx_buf(adapter, rx_ring, cleaned_count); cleaned_count = 0; + xdp_xmit = 0; } /* use prefetched values */ @@ -4498,8 +4536,11 @@ next_desc: rx_ring->next_to_clean = i; cleaned_count = E1000_DESC_UNUSED(rx_ring); - if (cleaned_count) + if (cleaned_count) { + if (xdp_xmit) + e1000_xdp_xmit_bundle(xdp_bundle, netdev, adapter); adapter->alloc_rx_buf(adapter, rx_ring, cleaned_count); + } adapter->total_rx_packets += total_rx_packets; adapter->total_rx_bytes += total_rx_bytes;