From: Andy Gospodarek <gospo@redhat.com>
Date: Tue, 9 Sep 2008 18:50:25 -0400
Subject: [net] bnx2x: update to upstream version 1.45.21
Message-id: 20080909225025.GO5762@gospo.rdu.redhat.com
O-Subject: [RHEL5.3 PATCH] bnx2x: update to upstream version 1.45.21
Bugzilla: 442026
RH-Acked-by: David Miller <davem@redhat.com>

This is an update to the bnx2x patch I posted earlier.  A problem was
discovered where the buffer that was allocated was larger than the
PCI-mapped size.  This should really only be a problem on systems with
64k pages (PPC), but still important.  The version number was also
bumped up to 1.45.21.  It comes from the following upstream commit:

commit 437cf2f1c5f05e34510f43e129af29a716b04ce6
Author: Eilon Greenstein <eilong@broadcom.com>
Date:   Wed Sep 3 14:38:00 2008 -0700

    bnx2x: Accessing un-mapped page

    The allocated RX buffer size was 64 bytes bigger than the PCI mapped
    size with no good reason. If the packet was actually using the buffer up
    to its limit and if the last 64 bytes of the buffer crossed 4KB boundary
    then an unmapped PCI page was accessed. The fix is to use only one
    parameter for the buffer size - there is no need to differentiate
    between the buffer size and the PCI mapping size since the extra 64
    bytes can actually be used by the FW to align the Ethernet payload to
    64 bytes.

    Also updating the driver version and date

    Signed-off-by: Eilon Greenstein <eilong@broadcom.com>
    Signed-off-by: David S. Miller <davem@davemloft.net>

It has been tested by Broadcom and resolves the issue from 442026.

diff --git a/drivers/net/bnx2x.h b/drivers/net/bnx2x.h
index 0427f34..9594f71 100644
--- a/drivers/net/bnx2x.h
+++ b/drivers/net/bnx2x.h
@@ -151,6 +151,8 @@ struct sw_rx_page {
 #define PAGES_PER_SGE_SHIFT		0
 #define PAGES_PER_SGE			(1 << PAGES_PER_SGE_SHIFT)
 
+#define BCM_RX_ETH_PAYLOAD_ALIGN	64
+
 /* SGE ring related macros */
 #define NUM_RX_SGE_PAGES		2
 #define RX_SGE_CNT		(BCM_PAGE_SIZE / sizeof(struct eth_rx_sge))
@@ -748,8 +750,7 @@ struct bnx2x {
 
 	u32			rx_csum;
 	u32			rx_offset;
-	u32			rx_buf_use_size;	/* useable size */
-	u32			rx_buf_size;		/* with alignment */
+	u32			rx_buf_size;
 #define ETH_OVREHEAD			(ETH_HLEN + 8)	/* 8 for CRC + VLAN */
 #define ETH_MIN_PACKET_SIZE		60
 #define ETH_MAX_PACKET_SIZE		1500
diff --git a/drivers/net/bnx2x_main.c b/drivers/net/bnx2x_main.c
index 319c2b0..b01eee3 100644
--- a/drivers/net/bnx2x_main.c
+++ b/drivers/net/bnx2x_main.c
@@ -59,8 +59,8 @@
 #include "bnx2x.h"
 #include "bnx2x_init.h"
 
-#define DRV_MODULE_VERSION	"1.45.20"
-#define DRV_MODULE_RELDATE	"2008/08/25"
+#define DRV_MODULE_VERSION	"1.45.21"
+#define DRV_MODULE_RELDATE	"2008/09/03"
 #define BNX2X_BC_VER		0x040200
 
 /* Time in jiffies before concluding the transmitter is hung */
@@ -1029,7 +1029,7 @@ static inline int bnx2x_alloc_rx_skb(struct bnx2x *bp,
 	if (unlikely(skb == NULL))
 		return -ENOMEM;
 
-	mapping = pci_map_single(bp->pdev, skb->data, bp->rx_buf_use_size,
+	mapping = pci_map_single(bp->pdev, skb->data, bp->rx_buf_size,
 				 PCI_DMA_FROMDEVICE);
 	if (unlikely(dma_mapping_error(mapping))) {
 		dev_kfree_skb(skb);
@@ -1171,7 +1171,7 @@ static void bnx2x_tpa_start(struct bnx2x_fastpath *fp, u16 queue,
 	/* move empty skb from pool to prod and map it */
 	prod_rx_buf->skb = fp->tpa_pool[queue].skb;
 	mapping = pci_map_single(bp->pdev, fp->tpa_pool[queue].skb->data,
-				 bp->rx_buf_use_size, PCI_DMA_FROMDEVICE);
+				 bp->rx_buf_size, PCI_DMA_FROMDEVICE);
 	pci_unmap_addr_set(prod_rx_buf, mapping, mapping);
 
 	/* move partial skb from cons to pool (don't unmap yet) */
@@ -1278,7 +1278,7 @@ static void bnx2x_tpa_stop(struct bnx2x *bp, struct bnx2x_fastpath *fp,
 	   pool entry status to BNX2X_TPA_STOP even if new skb allocation
 	   fails. */
 	pci_unmap_single(bp->pdev, pci_unmap_addr(rx_buf, mapping),
-			 bp->rx_buf_use_size, PCI_DMA_FROMDEVICE);
+			 bp->rx_buf_size, PCI_DMA_FROMDEVICE);
 
 	if (likely(new_skb)) {
 		/* fix ip xsum and give it to the stack */
@@ -1522,7 +1522,7 @@ static int bnx2x_rx_int(struct bnx2x_fastpath *fp, int budget)
 			} else if (bnx2x_alloc_rx_skb(bp, fp, bd_prod) == 0) {
 				pci_unmap_single(bp->pdev,
 					pci_unmap_addr(rx_buf, mapping),
-						 bp->rx_buf_use_size,
+						 bp->rx_buf_size,
 						 PCI_DMA_FROMDEVICE);
 				skb_reserve(skb, pad);
 				skb_put(skb, len);
@@ -4234,7 +4234,7 @@ static inline void bnx2x_free_tpa_pool(struct bnx2x *bp,
 		if (fp->tpa_state[i] == BNX2X_TPA_START)
 			pci_unmap_single(bp->pdev,
 					 pci_unmap_addr(rx_buf, mapping),
-					 bp->rx_buf_use_size,
+					 bp->rx_buf_size,
 					 PCI_DMA_FROMDEVICE);
 
 		dev_kfree_skb(skb);
@@ -4250,15 +4250,14 @@ static void bnx2x_init_rx_rings(struct bnx2x *bp)
 	u16 ring_prod, cqe_ring_prod;
 	int i, j;
 
-	bp->rx_buf_use_size = bp->dev->mtu;
-	bp->rx_buf_use_size += bp->rx_offset + ETH_OVREHEAD;
-	bp->rx_buf_size = bp->rx_buf_use_size + 64;
+	bp->rx_buf_size = bp->dev->mtu;
+	bp->rx_buf_size += bp->rx_offset + ETH_OVREHEAD +
+		BCM_RX_ETH_PAYLOAD_ALIGN;
 
 	if (bp->flags & TPA_ENABLE_FLAG) {
 		DP(NETIF_MSG_IFUP,
-		   "rx_buf_use_size %d  rx_buf_size %d  effective_mtu %d\n",
-		   bp->rx_buf_use_size, bp->rx_buf_size,
-		   bp->dev->mtu + ETH_OVREHEAD);
+		   "rx_buf_size %d  effective_mtu %d\n",
+		   bp->rx_buf_size, bp->dev->mtu + ETH_OVREHEAD);
 
 		for_each_queue(bp, j) {
 			struct bnx2x_fastpath *fp = &bp->fp[j];
@@ -4467,9 +4466,10 @@ static void bnx2x_init_context(struct bnx2x *bp)
 		context->ustorm_st_context.common.status_block_id = sb_id;
 		context->ustorm_st_context.common.flags =
 			USTORM_ETH_ST_CONTEXT_CONFIG_ENABLE_MC_ALIGNMENT;
-		context->ustorm_st_context.common.mc_alignment_size = 64;
+		context->ustorm_st_context.common.mc_alignment_size =
+			BCM_RX_ETH_PAYLOAD_ALIGN;
 		context->ustorm_st_context.common.bd_buff_size =
-						bp->rx_buf_use_size;
+						bp->rx_buf_size;
 		context->ustorm_st_context.common.bd_page_base_hi =
 						U64_HI(fp->rx_desc_mapping);
 		context->ustorm_st_context.common.bd_page_base_lo =
@@ -4722,7 +4722,7 @@ static void bnx2x_init_internal_func(struct bnx2x *bp)
 	}
 
 	/* Init CQ ring mapping and aggregation size */
-	max_agg_size = min((u32)(bp->rx_buf_use_size +
+	max_agg_size = min((u32)(bp->rx_buf_size +
 				 8*BCM_PAGE_SIZE*PAGES_PER_SGE),
 			   (u32)0xffff);
 	for_each_queue(bp, i) {
@@ -5946,7 +5946,7 @@ static void bnx2x_free_rx_skbs(struct bnx2x *bp)
 
 			pci_unmap_single(bp->pdev,
 					 pci_unmap_addr(rx_buf, mapping),
-					 bp->rx_buf_use_size,
+					 bp->rx_buf_size,
 					 PCI_DMA_FROMDEVICE);
 
 			rx_buf->skb = NULL;