netdev - [PATCH net-next 12/18] net: mvpp2: improve the distribution of packets on CPUs when using RSS

lists.openwall.net		lists / announce owl-users owl-dev john-users john-dev passwdqc-users yescrypt popa3d-users / oss-security kernel-hardening musl sabotage tlsify passwords / crypt-dev xvendor / Bugtraq Full-Disclosure linux-kernel linux-netdev linux-ext4 linux-hardening linux-cve-announce PHC
Open Source and information security mailing list archives

Hash Suite: Windows password security audit tool. GUI, reports in PDF.

[<prev] [next>] [<thread-prev] [thread-next>] [day] [month] [year] [list]

Date:   Thu, 12 Jul 2018 13:54:21 +0200
From:   Maxime Chevallier <maxime.chevallier@...tlin.com>
To:     davem@...emloft.net
Cc:     Antoine Tenart <antoine.tenart@...tlin.com>,
        netdev@...r.kernel.org, linux-kernel@...r.kernel.org,
        thomas.petazzoni@...tlin.com, gregory.clement@...tlin.com,
        miquel.raynal@...tlin.com, nadavh@...vell.com, stefanc@...vell.com,
        ymarkman@...vell.com, mw@...ihalf.com,
        Maxime Chevallier <maxime.chevallier@...tlin.com>
Subject: [PATCH net-next 12/18] net: mvpp2: improve the distribution of packets on CPUs when using RSS

From: Antoine Tenart <antoine.tenart@...tlin.com>

This patch adds an extra indirection when setting the indirection table
into the RSS hardware table to improve the packets distribution across
CPUs. For example, if 2 queues are used on a multi-core system this new
indirection will choose two queues on two different CPUs instead of the
two first queues which are on the same first CPU.

Signed-off-by: Antoine Tenart <antoine.tenart@...tlin.com>
Signed-off-by: Maxime Chevallier <maxime.chevallier@...tlin.com>
---
 drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c | 16 +++++++++++++++-
 1 file changed, 15 insertions(+), 1 deletion(-)

diff --git a/drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c b/drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c
index c80a1a549224..7dafc8c425b8 100644
--- a/drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c
+++ b/drivers/net/ethernet/marvell/mvpp2/mvpp2_cls.c
@@ -107,6 +107,19 @@ void mvpp2_cls_oversize_rxq_set(struct mvpp2_port *port)
 	mvpp2_write(port->priv, MVPP2_CLS_SWFWD_PCTRL_REG, val);
 }
 
+static inline u32 mvpp22_rxfh_indir(struct mvpp2_port *port, u32 rxq)
+{
+	int nrxqs, cpus = num_possible_cpus();
+
+	/* Number of RXQs per CPU */
+	nrxqs = port->nrxqs / cpus;
+
+	/* Indirection to better distribute the paquets on the CPUs when
+	 * configuring the RSS queues.
+	 */
+	return port->first_rxq + ((rxq * nrxqs + rxq / cpus) % port->nrxqs);
+}
+
 void mvpp22_rss_fill_table(struct mvpp2_port *port, u32 table)
 {
 	struct mvpp2 *priv = port->priv;
@@ -117,7 +130,8 @@ void mvpp22_rss_fill_table(struct mvpp2_port *port, u32 table)
 			  MVPP22_RSS_INDEX_TABLE_ENTRY(i);
 		mvpp2_write(priv, MVPP22_RSS_INDEX, sel);
 
-		mvpp2_write(priv, MVPP22_RSS_TABLE_ENTRY, port->indir[i]);
+		mvpp2_write(priv, MVPP22_RSS_TABLE_ENTRY,
+			    mvpp22_rxfh_indir(port, port->indir[i]));
 	}
 }
 
-- 
2.11.0