From patchwork Tue Apr 30 12:06:34 2024 Content-Type: text/plain; charset="utf-8" MIME-Version: 1.0 Content-Transfer-Encoding: 7bit X-Patchwork-Submitter: MD Danish Anwar X-Patchwork-Id: 13648896 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from bombadil.infradead.org (bombadil.infradead.org [198.137.202.133]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-GCM-SHA384 (256/256 bits)) (No client certificate requested) by smtp.lore.kernel.org (Postfix) with ESMTPS id A2214C4345F for ; Tue, 30 Apr 2024 12:07:06 +0000 (UTC) DKIM-Signature: v=1; a=rsa-sha256; q=dns/txt; c=relaxed/relaxed; d=lists.infradead.org; s=bombadil.20210309; h=Sender: Content-Transfer-Encoding:Content-Type:List-Subscribe:List-Help:List-Post: List-Archive:List-Unsubscribe:List-Id:MIME-Version:Message-ID:Date:Subject:CC :To:From:Reply-To:Content-ID:Content-Description:Resent-Date:Resent-From: Resent-Sender:Resent-To:Resent-Cc:Resent-Message-ID:In-Reply-To:References: List-Owner; bh=m5+1fvvOqScO+mlz4Tf5gLQrh4jmcO5TAbLT4/fpn5Q=; b=1K5PNi0eybRg6e vJqQ2OZSZJKJdkhr3KfClLK3c7VHTwH+VpkdrpbYHIwqoOkqUfos170dTg2grHlemP98M+8JfmG/Q owhvfVJA8Ud8uvqWMyfNprJMlTVVKNBZppiLuuY+vyW5dsNNhMUfduqAzsiKVLiPwyIboofTe6rQA QY9a+WHKe0fLVp5KAVo59oH9UdjTE7MNwZNJflTJk9HcBjrHJICjsm6ipW/CjZuimL2DGUV2TkYVz GudP0ZG12aG1AbZdlQcNpMrkRSynuOZ2EUeudrR37+KBH5jUSFUZy9Y1L/knGV2Cu7n8Zjf+kDnWc uEXM3BwQD/Az/E+NsTKA==; Received: from localhost ([::1] helo=bombadil.infradead.org) by bombadil.infradead.org with esmtp (Exim 4.97.1 #2 (Red Hat Linux)) id 1s1mG5-00000006Hno-39s6; Tue, 30 Apr 2024 12:06:53 +0000 Received: from fllv0016.ext.ti.com ([198.47.19.142]) by bombadil.infradead.org with esmtps (Exim 4.97.1 #2 (Red Hat Linux)) id 1s1mG2-00000006Hmg-0K0p for linux-arm-kernel@lists.infradead.org; Tue, 30 Apr 2024 12:06:51 +0000 Received: from fllv0034.itg.ti.com ([10.64.40.246]) by fllv0016.ext.ti.com (8.15.2/8.15.2) with ESMTP id 43UC6c2S019884; Tue, 30 Apr 2024 07:06:38 -0500 DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed; d=ti.com; s=ti-com-17Q1; t=1714478798; bh=tcUCvNxzuyXMP7T6wVE8YtL/6igb/ha5hCDb+AUyyKQ=; h=From:To:CC:Subject:Date; b=AoXsyqUymYdSpWu16DC54hITwlfo6UjAOxsrf9HyXGq7LTVbVucJxU3bC4BBfHjxo 1pC1W6+o5TcIH5MKHEznAvrfEy7JH91KppACGFJ19pdHVkb5r5ElETe3jQ7zkA2YXT RGTG5MhyzAFHVQGILH7dy6Kr5b4kLvFaA7gq3gLU= Received: from DLEE115.ent.ti.com (dlee115.ent.ti.com [157.170.170.26]) by fllv0034.itg.ti.com (8.15.2/8.15.2) with ESMTPS id 43UC6cPt082146 (version=TLSv1.2 cipher=AES256-GCM-SHA384 bits=256 verify=FAIL); Tue, 30 Apr 2024 07:06:38 -0500 Received: from DLEE114.ent.ti.com (157.170.170.25) by DLEE115.ent.ti.com (157.170.170.26) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256_P256) id 15.1.2507.23; Tue, 30 Apr 2024 07:06:37 -0500 Received: from lelvsmtp6.itg.ti.com (10.180.75.249) by DLEE114.ent.ti.com (157.170.170.25) with Microsoft SMTP Server (version=TLS1_2, cipher=TLS_ECDHE_RSA_WITH_AES_128_CBC_SHA256_P256) id 15.1.2507.23 via Frontend Transport; Tue, 30 Apr 2024 07:06:37 -0500 Received: from lelv0854.itg.ti.com (lelv0854.itg.ti.com [10.181.64.140]) by lelvsmtp6.itg.ti.com (8.15.2/8.15.2) with ESMTP id 43UC6bhf043702; Tue, 30 Apr 2024 07:06:37 -0500 Received: from localhost (danish-tpc.dhcp.ti.com [10.24.69.25]) by lelv0854.itg.ti.com (8.14.7/8.14.7) with ESMTP id 43UC6a7C024808; Tue, 30 Apr 2024 07:06:36 -0500 From: MD Danish Anwar To: Dan Carpenter , Heiner Kallweit , Andrew Lunn , Jan Kiszka , Diogo Ivo , Paolo Abeni , Jakub Kicinski , Eric Dumazet , "David S. Miller" CC: , , , , Vignesh Raghavendra , , Roger Quadros , Simon Horman , MD Danish Anwar Subject: [PATCH net-next v3] net: ti: icssg_prueth: Add SW TX / RX Coalescing based on hrtimers Date: Tue, 30 Apr 2024 17:36:34 +0530 Message-ID: <20240430120634.1558998-1-danishanwar@ti.com> X-Mailer: git-send-email 2.34.1 MIME-Version: 1.0 X-EXCLAIMER-MD-CONFIG: e1e8a2fd-e40a-4ac6-ac9b-f7e9cc9ee180 X-CRM114-Version: 20100106-BlameMichelson ( TRE 0.8.0 (BSD) ) MR-646709E3 X-CRM114-CacheID: sfid-20240430_050650_309542_F8222E48 X-CRM114-Status: GOOD ( 21.71 ) X-BeenThere: linux-arm-kernel@lists.infradead.org X-Mailman-Version: 2.1.34 Precedence: list List-Id: List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: "linux-arm-kernel" Errors-To: linux-arm-kernel-bounces+linux-arm-kernel=archiver.kernel.org@lists.infradead.org Add SW IRQ coalescing based on hrtimers for RX and TX data path for ICSSG driver, which can be enabled by ethtool commands: - RX coalescing ethtool -C eth1 rx-usecs 50 - TX coalescing can be enabled per TX queue - by default enables coalescing for TX0 ethtool -C eth1 tx-usecs 50 - configure TX0 ethtool -Q eth0 queue_mask 1 --coalesce tx-usecs 100 - configure TX1 ethtool -Q eth0 queue_mask 2 --coalesce tx-usecs 100 - configure TX0 and TX1 ethtool -Q eth0 queue_mask 3 --coalesce tx-usecs 100 --coalesce tx-usecs 100 Minimum value for both rx-usecs and tx-usecs is 20us. Compared to gro_flush_timeout and napi_defer_hard_irqs this patch allows to enable IRQ coalescing for RX path separately. Benchmarking numbers: =============================================================== | Method | Tput_TX | CPU_TX | Tput_RX | CPU_RX | | ============================================================== | Default Driver 943 Mbps 31% 517 Mbps 38% | | IRQ Coalescing (Patch) 943 Mbps 28% 518 Mbps 25% | =============================================================== Signed-off-by: MD Danish Anwar Reviewed-by: Andrew Lunn Reviewed-by: Simon Horman --- Changes from v2 [2] to v3: *) Collected RB tag from Andrew Lunn *) Fixed warning caused by {} missing for outer if conditions in emac_napi_tx_poll() and emac_napi_rx_poll() APIs. *) Fixed spelling mistake. Changes from v1 [1] to v2: *) Added Benchmarking numbers in the commit message as suggested by Andrew Lunn . Full logs [3] *) Addressed comments given by Simon Horman in v1. [1] https://lore.kernel.org/all/20240424091823.1814136-1-danishanwar@ti.com/ [2] https://lore.kernel.org/all/20240429071501.547680-1-danishanwar@ti.com/ [3] https://gist.githubusercontent.com/danish-ti/47855631be9f3635cee994693662a988/raw/94b4eb86b42fe243ab03186a88a314e0cb272fd0/gistfile1.txt drivers/net/ethernet/ti/icssg/icssg_common.c | 41 ++++++-- drivers/net/ethernet/ti/icssg/icssg_ethtool.c | 93 +++++++++++++++++++ drivers/net/ethernet/ti/icssg/icssg_prueth.c | 18 +++- drivers/net/ethernet/ti/icssg/icssg_prueth.h | 11 ++- 4 files changed, 155 insertions(+), 8 deletions(-) base-commit: 9f02bb6d7a229058ffaba4f6dd78e0f7b06b799c diff --git a/drivers/net/ethernet/ti/icssg/icssg_common.c b/drivers/net/ethernet/ti/icssg/icssg_common.c index 284f97876054..088ab8076db4 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_common.c +++ b/drivers/net/ethernet/ti/icssg/icssg_common.c @@ -122,7 +122,7 @@ void prueth_xmit_free(struct prueth_tx_chn *tx_chn, } int emac_tx_complete_packets(struct prueth_emac *emac, int chn, - int budget) + int budget, bool *tdown) { struct net_device *ndev = emac->ndev; struct cppi5_host_desc_t *desc_tx; @@ -145,6 +145,7 @@ int emac_tx_complete_packets(struct prueth_emac *emac, int chn, if (cppi5_desc_is_tdcm(desc_dma)) { if (atomic_dec_and_test(&emac->tdown_cnt)) complete(&emac->tdown_complete); + *tdown = true; break; } @@ -190,19 +191,37 @@ int emac_tx_complete_packets(struct prueth_emac *emac, int chn, return num_tx; } +static enum hrtimer_restart emac_tx_timer_callback(struct hrtimer *timer) +{ + struct prueth_tx_chn *tx_chns = + container_of(timer, struct prueth_tx_chn, tx_hrtimer); + + enable_irq(tx_chns->irq); + return HRTIMER_NORESTART; +} + static int emac_napi_tx_poll(struct napi_struct *napi_tx, int budget) { struct prueth_tx_chn *tx_chn = prueth_napi_to_tx_chn(napi_tx); struct prueth_emac *emac = tx_chn->emac; + bool tdown = false; int num_tx_packets; - num_tx_packets = emac_tx_complete_packets(emac, tx_chn->id, budget); + num_tx_packets = emac_tx_complete_packets(emac, tx_chn->id, budget, + &tdown); if (num_tx_packets >= budget) return budget; - if (napi_complete_done(napi_tx, num_tx_packets)) - enable_irq(tx_chn->irq); + if (napi_complete_done(napi_tx, num_tx_packets)) { + if (unlikely(tx_chn->tx_pace_timeout_ns && !tdown)) { + hrtimer_start(&tx_chn->tx_hrtimer, + ns_to_ktime(tx_chn->tx_pace_timeout_ns), + HRTIMER_MODE_REL_PINNED); + } else { + enable_irq(tx_chn->irq); + } + } return num_tx_packets; } @@ -226,6 +245,9 @@ int prueth_ndev_add_tx_napi(struct prueth_emac *emac) struct prueth_tx_chn *tx_chn = &emac->tx_chns[i]; netif_napi_add_tx(emac->ndev, &tx_chn->napi_tx, emac_napi_tx_poll); + hrtimer_init(&tx_chn->tx_hrtimer, CLOCK_MONOTONIC, + HRTIMER_MODE_REL_PINNED); + tx_chn->tx_hrtimer.function = &emac_tx_timer_callback; ret = request_irq(tx_chn->irq, prueth_tx_irq, IRQF_TRIGGER_HIGH, tx_chn->name, tx_chn); @@ -871,8 +893,15 @@ int emac_napi_rx_poll(struct napi_struct *napi_rx, int budget) break; } - if (num_rx < budget && napi_complete_done(napi_rx, num_rx)) - enable_irq(emac->rx_chns.irq[rx_flow]); + if (num_rx < budget && napi_complete_done(napi_rx, num_rx)) { + if (unlikely(emac->rx_pace_timeout_ns)) { + hrtimer_start(&emac->rx_hrtimer, + ns_to_ktime(emac->rx_pace_timeout_ns), + HRTIMER_MODE_REL_PINNED); + } else { + enable_irq(emac->rx_chns.irq[rx_flow]); + } + } return num_rx; } diff --git a/drivers/net/ethernet/ti/icssg/icssg_ethtool.c b/drivers/net/ethernet/ti/icssg/icssg_ethtool.c index ca20325d4d3e..c8d0f45cc5b1 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_ethtool.c +++ b/drivers/net/ethernet/ti/icssg/icssg_ethtool.c @@ -201,6 +201,93 @@ static void emac_get_rmon_stats(struct net_device *ndev, rmon_stats->hist_tx[4] = emac_get_stat_by_name(emac, "tx_bucket5_frames"); } +static int emac_get_coalesce(struct net_device *ndev, + struct ethtool_coalesce *coal, + struct kernel_ethtool_coalesce *kernel_coal, + struct netlink_ext_ack *extack) +{ + struct prueth_emac *emac = netdev_priv(ndev); + struct prueth_tx_chn *tx_chn; + + tx_chn = &emac->tx_chns[0]; + + coal->rx_coalesce_usecs = emac->rx_pace_timeout_ns / 1000; + coal->tx_coalesce_usecs = tx_chn->tx_pace_timeout_ns / 1000; + + return 0; +} + +static int emac_get_per_queue_coalesce(struct net_device *ndev, u32 queue, + struct ethtool_coalesce *coal) +{ + struct prueth_emac *emac = netdev_priv(ndev); + struct prueth_tx_chn *tx_chn; + + if (queue >= PRUETH_MAX_TX_QUEUES) + return -EINVAL; + + tx_chn = &emac->tx_chns[queue]; + + coal->tx_coalesce_usecs = tx_chn->tx_pace_timeout_ns / 1000; + + return 0; +} + +static int emac_set_coalesce(struct net_device *ndev, + struct ethtool_coalesce *coal, + struct kernel_ethtool_coalesce *kernel_coal, + struct netlink_ext_ack *extack) +{ + struct prueth_emac *emac = netdev_priv(ndev); + struct prueth *prueth = emac->prueth; + struct prueth_tx_chn *tx_chn; + + tx_chn = &emac->tx_chns[0]; + + if (coal->rx_coalesce_usecs && + coal->rx_coalesce_usecs < ICSSG_MIN_COALESCE_USECS) { + dev_info(prueth->dev, "defaulting to min value of %dus for rx-usecs\n", + ICSSG_MIN_COALESCE_USECS); + coal->rx_coalesce_usecs = ICSSG_MIN_COALESCE_USECS; + } + + if (coal->tx_coalesce_usecs && + coal->tx_coalesce_usecs < ICSSG_MIN_COALESCE_USECS) { + dev_info(prueth->dev, "defaulting to min value of %dus for tx-usecs\n", + ICSSG_MIN_COALESCE_USECS); + coal->tx_coalesce_usecs = ICSSG_MIN_COALESCE_USECS; + } + + emac->rx_pace_timeout_ns = coal->rx_coalesce_usecs * 1000; + tx_chn->tx_pace_timeout_ns = coal->tx_coalesce_usecs * 1000; + + return 0; +} + +static int emac_set_per_queue_coalesce(struct net_device *ndev, u32 queue, + struct ethtool_coalesce *coal) +{ + struct prueth_emac *emac = netdev_priv(ndev); + struct prueth *prueth = emac->prueth; + struct prueth_tx_chn *tx_chn; + + if (queue >= PRUETH_MAX_TX_QUEUES) + return -EINVAL; + + tx_chn = &emac->tx_chns[queue]; + + if (coal->tx_coalesce_usecs && + coal->tx_coalesce_usecs < ICSSG_MIN_COALESCE_USECS) { + dev_info(prueth->dev, "defaulting to min value of %dus for tx-usecs for tx-%u\n", + ICSSG_MIN_COALESCE_USECS, queue); + coal->tx_coalesce_usecs = ICSSG_MIN_COALESCE_USECS; + } + + tx_chn->tx_pace_timeout_ns = coal->tx_coalesce_usecs * 1000; + + return 0; +} + const struct ethtool_ops icssg_ethtool_ops = { .get_drvinfo = emac_get_drvinfo, .get_msglevel = emac_get_msglevel, @@ -209,6 +296,12 @@ const struct ethtool_ops icssg_ethtool_ops = { .get_ethtool_stats = emac_get_ethtool_stats, .get_strings = emac_get_strings, .get_ts_info = emac_get_ts_info, + .supported_coalesce_params = ETHTOOL_COALESCE_RX_USECS | + ETHTOOL_COALESCE_TX_USECS, + .get_coalesce = emac_get_coalesce, + .set_coalesce = emac_set_coalesce, + .get_per_queue_coalesce = emac_get_per_queue_coalesce, + .set_per_queue_coalesce = emac_set_per_queue_coalesce, .get_channels = emac_get_channels, .set_channels = emac_set_channels, .get_link_ksettings = emac_get_link_ksettings, diff --git a/drivers/net/ethernet/ti/icssg/icssg_prueth.c b/drivers/net/ethernet/ti/icssg/icssg_prueth.c index 186b0365c2e5..7c9e9518f555 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_prueth.c +++ b/drivers/net/ethernet/ti/icssg/icssg_prueth.c @@ -243,6 +243,16 @@ static void emac_adjust_link(struct net_device *ndev) } } +static enum hrtimer_restart emac_rx_timer_callback(struct hrtimer *timer) +{ + struct prueth_emac *emac = + container_of(timer, struct prueth_emac, rx_hrtimer); + int rx_flow = PRUETH_RX_FLOW_DATA; + + enable_irq(emac->rx_chns.irq[rx_flow]); + return HRTIMER_NORESTART; +} + static int emac_phy_connect(struct prueth_emac *emac) { struct prueth *prueth = emac->prueth; @@ -582,8 +592,10 @@ static int emac_ndo_stop(struct net_device *ndev) netdev_err(ndev, "tx teardown timeout\n"); prueth_reset_tx_chan(emac, emac->tx_ch_num, true); - for (i = 0; i < emac->tx_ch_num; i++) + for (i = 0; i < emac->tx_ch_num; i++) { napi_disable(&emac->tx_chns[i].napi_tx); + hrtimer_cancel(&emac->tx_chns[i].tx_hrtimer); + } max_rx_flows = PRUETH_MAX_RX_FLOWS; k3_udma_glue_tdown_rx_chn(emac->rx_chns.rx_chn, true); @@ -591,6 +603,7 @@ static int emac_ndo_stop(struct net_device *ndev) prueth_reset_rx_chan(&emac->rx_chns, max_rx_flows, true); napi_disable(&emac->napi_rx); + hrtimer_cancel(&emac->rx_hrtimer); cancel_work_sync(&emac->rx_mode_work); @@ -801,6 +814,9 @@ static int prueth_netdev_init(struct prueth *prueth, ndev->features = ndev->hw_features; netif_napi_add(ndev, &emac->napi_rx, emac_napi_rx_poll); + hrtimer_init(&emac->rx_hrtimer, CLOCK_MONOTONIC, + HRTIMER_MODE_REL_PINNED); + emac->rx_hrtimer.function = &emac_rx_timer_callback; prueth->emac[mac] = emac; return 0; diff --git a/drivers/net/ethernet/ti/icssg/icssg_prueth.h b/drivers/net/ethernet/ti/icssg/icssg_prueth.h index 82e38ef5635b..a78c5eb75fb8 100644 --- a/drivers/net/ethernet/ti/icssg/icssg_prueth.h +++ b/drivers/net/ethernet/ti/icssg/icssg_prueth.h @@ -108,6 +108,8 @@ struct prueth_tx_chn { u32 descs_num; unsigned int irq; char name[32]; + struct hrtimer tx_hrtimer; + unsigned long tx_pace_timeout_ns; }; struct prueth_rx_chn { @@ -127,6 +129,9 @@ struct prueth_rx_chn { #define PRUETH_MAX_TX_TS_REQUESTS 50 /* Max simultaneous TX_TS requests */ +/* Minimum coalesce time in usecs for both Tx and Rx */ +#define ICSSG_MIN_COALESCE_USECS 20 + /* data for each emac port */ struct prueth_emac { bool is_sr1; @@ -183,6 +188,10 @@ struct prueth_emac { struct delayed_work stats_work; u64 stats[ICSSG_NUM_STATS]; + + /* RX IRQ Coalescing Related */ + struct hrtimer rx_hrtimer; + unsigned long rx_pace_timeout_ns; }; /** @@ -320,7 +329,7 @@ void prueth_ndev_del_tx_napi(struct prueth_emac *emac, int num); void prueth_xmit_free(struct prueth_tx_chn *tx_chn, struct cppi5_host_desc_t *desc); int emac_tx_complete_packets(struct prueth_emac *emac, int chn, - int budget); + int budget, bool *tdown); int prueth_ndev_add_tx_napi(struct prueth_emac *emac); int prueth_init_tx_chns(struct prueth_emac *emac); int prueth_init_rx_chns(struct prueth_emac *emac,