netdev - [PATCH 1/5] mv643xx_eth: avoid dropping tx lock during transmit reclaim

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>] [day] [month] [year] [list]

Message-ID: <20080914171345.GC28798@xi.wantstofly.org>
Date:	Sun, 14 Sep 2008 19:13:45 +0200
From:	Lennert Buytenhek <buytenh@...tstofly.org>
To:	netdev@...r.kernel.org
Cc:	Dale Farnsworth <dale@...nsworth.org>,
	Ashish Karkare <akarkare@...vell.com>,
	Nicolas Pitre <nico@....org>
Subject: [PATCH 1/5] mv643xx_eth: avoid dropping tx lock during transmit reclaim

By moving DMA unmapping during transmit reclaim back under the netif
tx lock, we avoid the situation where we read the DMA address and buffer
length from the descriptor under the lock and then not do anything with
that data after dropping the lock on platforms where the DMA unmapping
routines are all NOPs (which is the case on all ARM platforms that
mv643xx_eth is used on at least).

This saves two uncached reads, which makes a small but measurable
performance difference in routing benchmarks.

Signed-off-by: Lennert Buytenhek <buytenh@...vell.com>
---
 drivers/net/mv643xx_eth.c |   22 +++++++---------------
 1 files changed, 7 insertions(+), 15 deletions(-)

diff --git a/drivers/net/mv643xx_eth.c b/drivers/net/mv643xx_eth.c
index d653b5a..7410eca 100644
--- a/drivers/net/mv643xx_eth.c
+++ b/drivers/net/mv643xx_eth.c
@@ -866,8 +866,6 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
 		struct tx_desc *desc;
 		u32 cmd_sts;
 		struct sk_buff *skb;
-		dma_addr_t addr;
-		int count;
 
 		tx_index = txq->tx_used_desc;
 		desc = &txq->tx_desc_area[tx_index];
@@ -886,8 +884,6 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
 		reclaimed++;
 		txq->tx_desc_count--;
 
-		addr = desc->buf_ptr;
-		count = desc->byte_cnt;
 		skb = txq->tx_skb[tx_index];
 		txq->tx_skb[tx_index] = NULL;
 
@@ -896,20 +892,16 @@ static int txq_reclaim(struct tx_queue *txq, int budget, int force)
 			mp->dev->stats.tx_errors++;
 		}
 
-		/*
-		 * Drop tx queue lock while we free the skb.
-		 */
-		__netif_tx_unlock(nq);
-
-		if (cmd_sts & TX_FIRST_DESC)
-			dma_unmap_single(NULL, addr, count, DMA_TO_DEVICE);
-		else
-			dma_unmap_page(NULL, addr, count, DMA_TO_DEVICE);
+		if (cmd_sts & TX_FIRST_DESC) {
+			dma_unmap_single(NULL, desc->buf_ptr,
+					 desc->byte_cnt, DMA_TO_DEVICE);
+		} else {
+			dma_unmap_page(NULL, desc->buf_ptr,
+				       desc->byte_cnt, DMA_TO_DEVICE);
+		}
 
 		if (skb)
 			dev_kfree_skb(skb);
-
-		__netif_tx_lock(nq, smp_processor_id());
 	}
 
 	__netif_tx_unlock(nq);
-- 
1.5.6.4
--
To unsubscribe from this list: send the line "unsubscribe netdev" in
the body of a message to majordomo@...r.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html