netdev - [PATCH net-next v2 3/7] bnxt_en: Allocate page pool per numa node

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]

Message-Id: <20240402093753.331120-4-pavan.chebbi@broadcom.com>
Date: Tue,  2 Apr 2024 02:37:49 -0700
From: Pavan Chebbi <pavan.chebbi@...adcom.com>
To: michael.chan@...adcom.com
Cc: davem@...emloft.net,
	edumazet@...gle.com,
	gospo@...adcom.com,
	kuba@...nel.org,
	netdev@...r.kernel.org,
	pabeni@...hat.com,
	Somnath Kotur <somnath.kotur@...adcom.com>,
	Andy Gospodarek <andrew.gospodarek@...adcom.com>,
	Pavan Chebbi <pavan.chebbi@...adcom.com>
Subject: [PATCH net-next v2 3/7] bnxt_en: Allocate page pool per numa node

From: Somnath Kotur <somnath.kotur@...adcom.com>

Driver's Page Pool allocation code looks at the node local
to the PCIe device to determine where to allocate memory.
In scenarios where the core count per NUMA node is low (< default rings)
it makes sense to exhaust page pool allocations on
Node 0 first and then moving on to allocating page pools
for the remaining rings from Node 1.

With this patch, and the following configuration on the NIC
$ ethtool -L ens1f0np0 combined 16
(core count/node = 12, first 12 rings on node#0, last 4 rings node#1)
and traffic redirected to a ring on node#1 , we see a performance
improvement of ~20%

Signed-off-by: Somnath Kotur <somnath.kotur@...adcom.com>
Reviewed-by: Andy Gospodarek <andrew.gospodarek@...adcom.com>
Reviewed-by: Michael Chan <michael.chan@...adcom.com>
Signed-off-by: Pavan Chebbi <pavan.chebbi@...adcom.com>
---
 drivers/net/ethernet/broadcom/bnxt/bnxt.c | 15 +++++++++++----
 1 file changed, 11 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnxt/bnxt.c b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
index 44b9332c147e..9fca1dee6486 100644
--- a/drivers/net/ethernet/broadcom/bnxt/bnxt.c
+++ b/drivers/net/ethernet/broadcom/bnxt/bnxt.c
@@ -3559,14 +3559,15 @@ static void bnxt_free_rx_rings(struct bnxt *bp)
 }
 
 static int bnxt_alloc_rx_page_pool(struct bnxt *bp,
-				   struct bnxt_rx_ring_info *rxr)
+				   struct bnxt_rx_ring_info *rxr,
+				   int numa_node)
 {
 	struct page_pool_params pp = { 0 };
 
 	pp.pool_size = bp->rx_agg_ring_size;
 	if (BNXT_RX_PAGE_MODE(bp))
 		pp.pool_size += bp->rx_ring_size;
-	pp.nid = dev_to_node(&bp->pdev->dev);
+	pp.nid = numa_node;
 	pp.napi = &rxr->bnapi->napi;
 	pp.netdev = bp->dev;
 	pp.dev = &bp->pdev->dev;
@@ -3586,7 +3587,8 @@ static int bnxt_alloc_rx_page_pool(struct bnxt *bp,
 
 static int bnxt_alloc_rx_rings(struct bnxt *bp)
 {
-	int i, rc = 0, agg_rings = 0;
+	int numa_node = dev_to_node(&bp->pdev->dev);
+	int i, rc = 0, agg_rings = 0, cpu;
 
 	if (!bp->rx_ring)
 		return -ENOMEM;
@@ -3597,10 +3599,15 @@ static int bnxt_alloc_rx_rings(struct bnxt *bp)
 	for (i = 0; i < bp->rx_nr_rings; i++) {
 		struct bnxt_rx_ring_info *rxr = &bp->rx_ring[i];
 		struct bnxt_ring_struct *ring;
+		int cpu_node;
 
 		ring = &rxr->rx_ring_struct;
 
-		rc = bnxt_alloc_rx_page_pool(bp, rxr);
+		cpu = cpumask_local_spread(i, numa_node);
+		cpu_node = cpu_to_node(cpu);
+		netdev_dbg(bp->dev, "Allocating page pool for rx_ring[%d] on numa_node: %d\n",
+			   i, cpu_node);
+		rc = bnxt_alloc_rx_page_pool(bp, rxr, cpu_node);
 		if (rc)
 			return rc;
 
-- 
2.39.1


Download attachment "smime.p7s" of type "application/pkcs7-signature" (4209 bytes)