[PATCH] forcedeth: fix dma api mismatches

[PATCH] forcedeth: fix dma api mismatches

[Eric Dumazet]

David & Ayaz

Please find following patch, candidate for stable, but alas not tested
since I dont have the hardware. I hit this on a 2.6.30 kernel on a machine
I dont have access anymore.

Thank you

[PATCH] forcedeth: fix dma api mismatches

forcedeth doesnt use properly dma api in its tx completion path
and in nv_loopback_test()

pci_map_single() should be paired with pci_unmap_single()
pci_map_page() should be paired with pci_unmap_page()

forcedeth xmit path uses pci_map_single() & pci_map_page(),
but tx completion path only uses pci_unmap_single()

nv_loopback_test() uses pci_map_single() & pci_unmap_page()

Add a dma_single field in struct nv_skb_map, and
define a helper function nv_unmap_txskb

Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
CC: Ayaz Abdulla <aabdulla@nvidia.com>
---
 drivers/net/forcedeth.c |   46 ++++++++++++++++++++++----------------
 1 files changed, 27 insertions(+), 19 deletions(-)

diff --git a/drivers/net/forcedeth.c b/drivers/net/forcedeth.c
index b60a304..1094d29 100644
--- a/drivers/net/forcedeth.c
+++ b/drivers/net/forcedeth.c
@@ -719,7 +719,8 @@ static const struct register_test nv_registers_test[] = {
 struct nv_skb_map {
 	struct sk_buff *skb;
 	dma_addr_t dma;
-	unsigned int dma_len;
+	unsigned int dma_len:31;
+	unsigned int dma_single:1;
 	struct ring_desc_ex *first_tx_desc;
 	struct nv_skb_map *next_tx_ctx;
 };
@@ -1912,6 +1913,7 @@ static void nv_init_tx(struct net_device *dev)
 		np->tx_skb[i].skb = NULL;
 		np->tx_skb[i].dma = 0;
 		np->tx_skb[i].dma_len = 0;
+		np->tx_skb[i].dma_single = 0;
 		np->tx_skb[i].first_tx_desc = NULL;
 		np->tx_skb[i].next_tx_ctx = NULL;
 	}
@@ -1930,23 +1932,30 @@ static int nv_init_ring(struct net_device *dev)
 		return nv_alloc_rx_optimized(dev);
 }
 
-static int nv_release_txskb(struct net_device *dev, struct nv_skb_map* tx_skb)
+static void nv_unmap_txskb(struct fe_priv *np, struct nv_skb_map *tx_skb)
 {
-	struct fe_priv *np = netdev_priv(dev);
-
 	if (tx_skb->dma) {
-		pci_unmap_page(np->pci_dev, tx_skb->dma,
-			       tx_skb->dma_len,
-			       PCI_DMA_TODEVICE);
+		if (tx_skb->dma_single)
+			pci_unmap_single(np->pci_dev, tx_skb->dma,
+					 tx_skb->dma_len,
+					 PCI_DMA_TODEVICE);
+		else
+			pci_unmap_page(np->pci_dev, tx_skb->dma,
+				       tx_skb->dma_len,
+				       PCI_DMA_TODEVICE);
 		tx_skb->dma = 0;
 	}
+}
+
+static int nv_release_txskb(struct fe_priv *np, struct nv_skb_map *tx_skb)
+{
+	nv_unmap_txskb(np, tx_skb);
 	if (tx_skb->skb) {
 		dev_kfree_skb_any(tx_skb->skb);
 		tx_skb->skb = NULL;
 		return 1;
-	} else {
-		return 0;
 	}
+	return 0;
 }
 
 static void nv_drain_tx(struct net_device *dev)
@@ -1964,10 +1973,11 @@ static void nv_drain_tx(struct net_device *dev)
 			np->tx_ring.ex[i].bufhigh = 0;
 			np->tx_ring.ex[i].buflow = 0;
 		}
-		if (nv_release_txskb(dev, &np->tx_skb[i]))
+		if (nv_release_txskb(np, &np->tx_skb[i]))
 			dev->stats.tx_dropped++;
 		np->tx_skb[i].dma = 0;
 		np->tx_skb[i].dma_len = 0;
+		np->tx_skb[i].dma_single = 0;
 		np->tx_skb[i].first_tx_desc = NULL;
 		np->tx_skb[i].next_tx_ctx = NULL;
 	}
@@ -2171,6 +2181,7 @@ static int nv_start_xmit(struct sk_buff *skb, struct net_device *dev)
 		np->put_tx_ctx->dma = pci_map_single(np->pci_dev, skb->data + offset, bcnt,
 						PCI_DMA_TODEVICE);
 		np->put_tx_ctx->dma_len = bcnt;
+		np->put_tx_ctx->dma_single = 1;
 		put_tx->buf = cpu_to_le32(np->put_tx_ctx->dma);
 		put_tx->flaglen = cpu_to_le32((bcnt-1) | tx_flags);
 
@@ -2196,6 +2207,7 @@ static int nv_start_xmit(struct sk_buff *skb, struct net_device *dev)
 			np->put_tx_ctx->dma = pci_map_page(np->pci_dev, frag->page, frag->page_offset+offset, bcnt,
 							   PCI_DMA_TODEVICE);
 			np->put_tx_ctx->dma_len = bcnt;
+			np->put_tx_ctx->dma_single = 0;
 			put_tx->buf = cpu_to_le32(np->put_tx_ctx->dma);
 			put_tx->flaglen = cpu_to_le32((bcnt-1) | tx_flags);
 
@@ -2291,6 +2303,7 @@ static int nv_start_xmit_optimized(struct sk_buff *skb, struct net_device *dev)
 		np->put_tx_ctx->dma = pci_map_single(np->pci_dev, skb->data + offset, bcnt,
 						PCI_DMA_TODEVICE);
 		np->put_tx_ctx->dma_len = bcnt;
+		np->put_tx_ctx->dma_single = 1;
 		put_tx->bufhigh = cpu_to_le32(dma_high(np->put_tx_ctx->dma));
 		put_tx->buflow = cpu_to_le32(dma_low(np->put_tx_ctx->dma));
 		put_tx->flaglen = cpu_to_le32((bcnt-1) | tx_flags);
@@ -2317,6 +2330,7 @@ static int nv_start_xmit_optimized(struct sk_buff *skb, struct net_device *dev)
 			np->put_tx_ctx->dma = pci_map_page(np->pci_dev, frag->page, frag->page_offset+offset, bcnt,
 							   PCI_DMA_TODEVICE);
 			np->put_tx_ctx->dma_len = bcnt;
+			np->put_tx_ctx->dma_single = 0;
 			put_tx->bufhigh = cpu_to_le32(dma_high(np->put_tx_ctx->dma));
 			put_tx->buflow = cpu_to_le32(dma_low(np->put_tx_ctx->dma));
 			put_tx->flaglen = cpu_to_le32((bcnt-1) | tx_flags);
@@ -2434,10 +2448,7 @@ static int nv_tx_done(struct net_device *dev, int limit)
 		dprintk(KERN_DEBUG "%s: nv_tx_done: flags 0x%x.\n",
 					dev->name, flags);
 
-		pci_unmap_page(np->pci_dev, np->get_tx_ctx->dma,
-			       np->get_tx_ctx->dma_len,
-			       PCI_DMA_TODEVICE);
-		np->get_tx_ctx->dma = 0;
+		nv_unmap_txskb(np, np->get_tx_ctx);
 
 		if (np->desc_ver == DESC_VER_1) {
 			if (flags & NV_TX_LASTPACKET) {
@@ -2502,10 +2513,7 @@ static int nv_tx_done_optimized(struct net_device *dev, int limit)
 		dprintk(KERN_DEBUG "%s: nv_tx_done_optimized: flags 0x%x.\n",
 					dev->name, flags);
 
-		pci_unmap_page(np->pci_dev, np->get_tx_ctx->dma,
-			       np->get_tx_ctx->dma_len,
-			       PCI_DMA_TODEVICE);
-		np->get_tx_ctx->dma = 0;
+		nv_unmap_txskb(np, np->get_tx_ctx);
 
 		if (flags & NV_TX2_LASTPACKET) {
 			if (!(flags & NV_TX2_ERROR))
@@ -5091,7 +5099,7 @@ static int nv_loopback_test(struct net_device *dev)
 		dprintk(KERN_DEBUG "%s: loopback - did not receive test packet\n", dev->name);
 	}
 
-	pci_unmap_page(np->pci_dev, test_dma_addr,
+	pci_unmap_single(np->pci_dev, test_dma_addr,
 		       (skb_end_pointer(tx_skb) - tx_skb->data),
 		       PCI_DMA_TODEVICE);
 	dev_kfree_skb_any(tx_skb);

Re: [PATCH] forcedeth: fix dma api mismatches

[David Miller]

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 18 Jun 2009 09:17:59 +0200

> Please find following patch, candidate for stable, but alas not tested
> since I dont have the hardware. I hit this on a 2.6.30 kernel on a machine
> I dont have access anymore.
> 
> Thank you
> 
> [PATCH] forcedeth: fix dma api mismatches
> 
> forcedeth doesnt use properly dma api in its tx completion path
> and in nv_loopback_test()
> 
> pci_map_single() should be paired with pci_unmap_single()
> pci_map_page() should be paired with pci_unmap_page()
> 
> forcedeth xmit path uses pci_map_single() & pci_map_page(),
> but tx completion path only uses pci_unmap_single()
> 
> nv_loopback_test() uses pci_map_single() & pci_unmap_page()
> 
> Add a dma_single field in struct nv_skb_map, and
> define a helper function nv_unmap_txskb
> 
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> CC: Ayaz Abdulla <aabdulla@nvidia.com>

I'm going to apply this to net-next-2.6 now since it looks
painfully correct to me.

RE: [PATCH] forcedeth: fix dma api mismatches

[Ayaz Abdulla]

Doing a quick indentifier search on 2.6.17 shows that pci_unmap_page() just calls pci_unmap_single().

I think this will just add extra clutter. We can just add a comment in the code instead that they perform the same thing.

Ayaz




-----Original Message-----
From: David Miller [mailto:davem@davemloft.net]
Sent: Thursday, June 18, 2009 12:47 AM
To: eric.dumazet@gmail.com
Cc: Ayaz Abdulla; netdev@vger.kernel.org
Subject: Re: [PATCH] forcedeth: fix dma api mismatches


From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Thu, 18 Jun 2009 09:17:59 +0200

> Please find following patch, candidate for stable, but alas not tested
> since I dont have the hardware. I hit this on a 2.6.30 kernel on a machine
> I dont have access anymore.
>
> Thank you
>
> [PATCH] forcedeth: fix dma api mismatches
>
> forcedeth doesnt use properly dma api in its tx completion path
> and in nv_loopback_test()
>
> pci_map_single() should be paired with pci_unmap_single()
> pci_map_page() should be paired with pci_unmap_page()
>
> forcedeth xmit path uses pci_map_single() & pci_map_page(),
> but tx completion path only uses pci_unmap_single()
>
> nv_loopback_test() uses pci_map_single() & pci_unmap_page()
>
> Add a dma_single field in struct nv_skb_map, and
> define a helper function nv_unmap_txskb
>
> Signed-off-by: Eric Dumazet <eric.dumazet@gmail.com>
> CC: Ayaz Abdulla <aabdulla@nvidia.com>

I'm going to apply this to net-next-2.6 now since it looks
painfully correct to me.
-----------------------------------------------------------------------------------
This email message is for the sole use of the intended recipient(s) and may contain
confidential information.  Any unauthorized review, use, disclosure or distribution
is prohibited.  If you are not the intended recipient, please contact the sender by
reply email and destroy all copies of the original message.
-----------------------------------------------------------------------------------

Re: [PATCH] forcedeth: fix dma api mismatches

[David Miller]

From: Ayaz Abdulla <AAbdulla@nvidia.com>
Date: Thu, 18 Jun 2009 10:33:43 -0700

> Doing a quick indentifier search on 2.6.17 shows that
> pci_unmap_page() just calls pci_unmap_single().

You should check more recent kernels.

> I think this will just add extra clutter. We can just add a comment
> in the code instead that they perform the same thing.

It causes log messages with the DMA API debugging layer that checks
whether you match up the calls correctly.