* [PATCH] fix natsemi PCI mapping
@ 2004-03-16 18:06 Jeff Garzik
2004-03-16 19:15 ` Steven J. Hill
0 siblings, 1 reply; 3+ messages in thread
From: Jeff Garzik @ 2004-03-16 18:06 UTC (permalink / raw)
To: Netdev, Tim Hockin; +Cc: Linux Kernel, ralf, sjhill
[-- Attachment #1: Type: text/plain, Size: 592 bytes --]
Somebody wanna review and/or test?
Changes:
* Define RX_OFFSET constant (value==2) for uses related to SKB
allocation and skb_reserve() calls
* RX skb's are always allocated maximally-sized, since we don't know the
size of an RX packet in advance. This means that we always alloc and
map RX skbs based on "np->rx_buf_sz + RX_OFFSET". natsemi got this
really wrong in refill_rx(), where it mapped skb->len just after
dev_alloc_skb(), which was very incorrect.
* call skb_reserve() in refill_rx(), our main skb allocation function,
just after dev_alloc_skb() returns successfully
[-- Attachment #2: patch --]
[-- Type: text/plain, Size: 3028 bytes --]
===== drivers/net/natsemi.c 1.58 vs edited =====
--- 1.58/drivers/net/natsemi.c Sun Mar 14 01:54:58 2004
+++ edited/drivers/net/natsemi.c Tue Mar 16 13:02:04 2004
@@ -175,6 +175,8 @@
#define DRV_VERSION "1.07+LK1.0.17"
#define DRV_RELDATE "Sep 27, 2002"
+#define RX_OFFSET 2
+
/* Updated to recommendations in pci-skeleton v2.03. */
/* The user-configurable values.
@@ -1467,13 +1469,16 @@
struct sk_buff *skb;
int entry = np->dirty_rx % RX_RING_SIZE;
if (np->rx_skbuff[entry] == NULL) {
- skb = dev_alloc_skb(np->rx_buf_sz);
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
+ skb = dev_alloc_skb(buflen);
np->rx_skbuff[entry] = skb;
if (skb == NULL)
break; /* Better luck next round. */
skb->dev = dev; /* Mark as being used by this device. */
+ /* 16 byte align the IP header */
+ skb_reserve(skb, RX_OFFSET);
np->rx_dma[entry] = pci_map_single(np->pci_dev,
- skb->data, skb->len, PCI_DMA_FROMDEVICE);
+ skb->tail, buflen, PCI_DMA_FROMDEVICE);
np->rx_ring[entry].addr = cpu_to_le32(np->rx_dma[entry]);
}
np->rx_ring[entry].cmd_status = cpu_to_le32(np->rx_buf_sz);
@@ -1543,6 +1548,7 @@
static void drain_ring(struct net_device *dev)
{
struct netdev_private *np = dev->priv;
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
int i;
/* Free all the skbuffs in the Rx queue. */
@@ -1551,7 +1557,7 @@
np->rx_ring[i].addr = 0xBADF00D0; /* An invalid address. */
if (np->rx_skbuff[i]) {
pci_unmap_single(np->pci_dev,
- np->rx_dma[i], np->rx_skbuff[i]->len,
+ np->rx_dma[i], buflen,
PCI_DMA_FROMDEVICE);
dev_kfree_skb(np->rx_skbuff[i]);
}
@@ -1747,6 +1753,7 @@
int entry = np->cur_rx % RX_RING_SIZE;
int boguscnt = np->dirty_rx + RX_RING_SIZE - np->cur_rx;
s32 desc_status = le32_to_cpu(np->rx_head_desc->cmd_status);
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
/* If the driver owns the next entry it's a new packet. Send it up. */
while (desc_status < 0) { /* e.g. & DescOwn */
@@ -1785,13 +1792,13 @@
/* Check if the packet is long enough to accept
* without copying to a minimally-sized skbuff. */
if (pkt_len < rx_copybreak
- && (skb = dev_alloc_skb(pkt_len + 2)) != NULL) {
+ && (skb = dev_alloc_skb(pkt_len + RX_OFFSET)) != NULL) {
skb->dev = dev;
/* 16 byte align the IP header */
- skb_reserve(skb, 2);
+ skb_reserve(skb, RX_OFFSET);
pci_dma_sync_single_for_cpu(np->pci_dev,
np->rx_dma[entry],
- np->rx_skbuff[entry]->len,
+ buflen,
PCI_DMA_FROMDEVICE);
#if HAS_IP_COPYSUM
eth_copy_and_sum(skb,
@@ -1803,12 +1810,11 @@
#endif
pci_dma_sync_single_for_device(np->pci_dev,
np->rx_dma[entry],
- np->rx_skbuff[entry]->len,
+ buflen,
PCI_DMA_FROMDEVICE);
} else {
pci_unmap_single(np->pci_dev, np->rx_dma[entry],
- np->rx_skbuff[entry]->len,
- PCI_DMA_FROMDEVICE);
+ buflen, PCI_DMA_FROMDEVICE);
skb_put(skb = np->rx_skbuff[entry], pkt_len);
np->rx_skbuff[entry] = NULL;
}
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] fix natsemi PCI mapping
2004-03-16 18:06 [PATCH] fix natsemi PCI mapping Jeff Garzik
@ 2004-03-16 19:15 ` Steven J. Hill
2004-03-16 19:20 ` Jeff Garzik
0 siblings, 1 reply; 3+ messages in thread
From: Steven J. Hill @ 2004-03-16 19:15 UTC (permalink / raw)
To: Jeff Garzik; +Cc: Netdev, Tim Hockin, Linux Kernel, ralf
[-- Attachment #1: Type: text/plain, Size: 1057 bytes --]
Jeff Garzik wrote:
>
> Somebody wanna review and/or test?
>
Hey Jeff.
I have tested this on 2.4 and it works great on MIPS with one
minor change below. Remove the 16 byte alignment of the IP
header. I discovered this when trying to do a BOOTP and mount
my NFS root filesystem. The BOOTP never succeeds. Patch against
latest 2.4.25 attached.
-Steve
@@ -1467,13 +1469,16 @@
struct sk_buff *skb;
int entry = np->dirty_rx % RX_RING_SIZE;
if (np->rx_skbuff[entry] == NULL) {
- skb = dev_alloc_skb(np->rx_buf_sz);
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
+ skb = dev_alloc_skb(buflen);
np->rx_skbuff[entry] = skb;
if (skb == NULL)
break; /* Better luck next round. */
skb->dev = dev; /* Mark as being used by this device. */
+ /* 16 byte align the IP header */
+ skb_reserve(skb, RX_OFFSET);
np->rx_dma[entry] = pci_map_single(np->pci_dev,
- skb->data, skb->len, PCI_DMA_FROMDEVICE);
+ skb->tail, buflen, PCI_DMA_FROMDEVICE);
np->rx_ring[entry].addr = cpu_to_le32(np->rx_dma[entry]);
}
[-- Attachment #2: natsemi-pci-mapping.patch --]
[-- Type: text/x-patch, Size: 2885 bytes --]
diff -urN -X /home/sjhill/diff-exc linux-2.4.25/drivers/net/natsemi.c linux-2.4.25-patched/drivers/net/natsemi.c
--- linux-2.4.25/drivers/net/natsemi.c Tue Mar 16 14:05:21 2004
+++ linux-2.4.25-patched/drivers/net/natsemi.c Tue Mar 16 13:58:15 2004
@@ -175,6 +175,8 @@
#define DRV_VERSION "1.07+LK1.0.17"
#define DRV_RELDATE "Sep 27, 2002"
+#define RX_OFFSET 2
+
/* Updated to recommendations in pci-skeleton v2.03. */
/* The user-configurable values.
@@ -1466,13 +1467,14 @@
struct sk_buff *skb;
int entry = np->dirty_rx % RX_RING_SIZE;
if (np->rx_skbuff[entry] == NULL) {
- skb = dev_alloc_skb(np->rx_buf_sz);
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
+ skb = dev_alloc_skb(buflen);
np->rx_skbuff[entry] = skb;
if (skb == NULL)
break; /* Better luck next round. */
skb->dev = dev; /* Mark as being used by this device. */
np->rx_dma[entry] = pci_map_single(np->pci_dev,
- skb->data, skb->len, PCI_DMA_FROMDEVICE);
+ skb->tail, buflen, PCI_DMA_FROMDEVICE);
np->rx_ring[entry].addr = cpu_to_le32(np->rx_dma[entry]);
}
np->rx_ring[entry].cmd_status = cpu_to_le32(np->rx_buf_sz);
@@ -1542,6 +1544,7 @@
static void drain_ring(struct net_device *dev)
{
struct netdev_private *np = dev->priv;
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
int i;
/* Free all the skbuffs in the Rx queue. */
@@ -1550,7 +1553,7 @@
np->rx_ring[i].addr = 0xBADF00D0; /* An invalid address. */
if (np->rx_skbuff[i]) {
pci_unmap_single(np->pci_dev,
- np->rx_dma[i], np->rx_skbuff[i]->len,
+ np->rx_dma[i], buflen,
PCI_DMA_FROMDEVICE);
dev_kfree_skb(np->rx_skbuff[i]);
}
@@ -1746,6 +1745,7 @@
int entry = np->cur_rx % RX_RING_SIZE;
int boguscnt = np->dirty_rx + RX_RING_SIZE - np->cur_rx;
s32 desc_status = le32_to_cpu(np->rx_head_desc->cmd_status);
+ unsigned int buflen = np->rx_buf_sz + RX_OFFSET;
/* If the driver owns the next entry it's a new packet. Send it up. */
while (desc_status < 0) { /* e.g. & DescOwn */
@@ -1784,13 +1784,13 @@
/* Check if the packet is long enough to accept
* without copying to a minimally-sized skbuff. */
if (pkt_len < rx_copybreak
- && (skb = dev_alloc_skb(pkt_len + 2)) != NULL) {
+ && (skb = dev_alloc_skb(pkt_len + RX_OFFSET)) != NULL) {
skb->dev = dev;
/* 16 byte align the IP header */
- skb_reserve(skb, 2);
+ skb_reserve(skb, RX_OFFSET);
pci_dma_sync_single(np->pci_dev,
np->rx_dma[entry],
- np->rx_skbuff[entry]->len,
+ buflen,
PCI_DMA_FROMDEVICE);
#if HAS_IP_COPYSUM
eth_copy_and_sum(skb,
@@ -1802,8 +1802,7 @@
#endif
} else {
pci_unmap_single(np->pci_dev, np->rx_dma[entry],
- np->rx_skbuff[entry]->len,
- PCI_DMA_FROMDEVICE);
+ buflen, PCI_DMA_FROMDEVICE);
skb_put(skb = np->rx_skbuff[entry], pkt_len);
np->rx_skbuff[entry] = NULL;
}
^ permalink raw reply [flat|nested] 3+ messages in thread
* Re: [PATCH] fix natsemi PCI mapping
2004-03-16 19:15 ` Steven J. Hill
@ 2004-03-16 19:20 ` Jeff Garzik
0 siblings, 0 replies; 3+ messages in thread
From: Jeff Garzik @ 2004-03-16 19:20 UTC (permalink / raw)
To: Steven J. Hill; +Cc: Netdev, Tim Hockin, Linux Kernel, ralf
Steven J. Hill wrote:
> Jeff Garzik wrote:
>
>>
>> Somebody wanna review and/or test?
>>
> Hey Jeff.
>
> I have tested this on 2.4 and it works great on MIPS with one
> minor change below. Remove the 16 byte alignment of the IP
> header. I discovered this when trying to do a BOOTP and mount
> my NFS root filesystem. The BOOTP never succeeds. Patch against
> latest 2.4.25 attached.
Interesting, thanks.
WRT skb_reserve(), I wonder if
(a) RX_OFFSET value passed to skb_reserve() needs to be per-arch
(b) the pre-existing skb_reserve() call is wrong too
^ permalink raw reply [flat|nested] 3+ messages in thread
end of thread, other threads:[~2004-03-16 19:20 UTC | newest]
Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2004-03-16 18:06 [PATCH] fix natsemi PCI mapping Jeff Garzik
2004-03-16 19:15 ` Steven J. Hill
2004-03-16 19:20 ` Jeff Garzik
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).