From: Brice Goglin <brice@myri.com>
To: Jeff Garzik <jeff@garzik.org>, netdev@vger.kernel.org
Subject: [PATCH 2/7] myri10ge: add page-based skb routines
Date: Mon, 11 Dec 2006 11:25:09 +0100 [thread overview]
Message-ID: <457D3205.6000203@myri.com> (raw)
In-Reply-To: <457D31B2.30505@myri.com>
Add physical page skb allocation routines and page based rx_done,
to be used by upcoming patches.
Signed-off-by: Brice Goglin <brice@myri.com>
---
drivers/net/myri10ge/myri10ge.c | 190 ++++++++++++++++++++++++++++++++++++++++
1 file changed, 190 insertions(+)
Index: linux-git/drivers/net/myri10ge/myri10ge.c
===================================================================
--- linux-git.orig/drivers/net/myri10ge/myri10ge.c 2006-12-11 10:52:08.000000000 +0100
+++ linux-git/drivers/net/myri10ge/myri10ge.c 2006-12-11 10:52:26.000000000 +0100
@@ -92,8 +92,14 @@
#define MYRI10GE_NO_CONFIRM_DATA htonl(0xffffffff)
#define MYRI10GE_NO_RESPONSE_RESULT 0xffffffff
+#define MYRI10GE_ALLOC_ORDER 0
+#define MYRI10GE_ALLOC_SIZE ((1 << MYRI10GE_ALLOC_ORDER) * PAGE_SIZE)
+#define MYRI10GE_MAX_FRAGS_PER_FRAME (MYRI10GE_MAX_ETHER_MTU/MYRI10GE_ALLOC_SIZE + 1)
+
struct myri10ge_rx_buffer_state {
struct sk_buff *skb;
+ struct page *page;
+ int page_offset;
DECLARE_PCI_UNMAP_ADDR(bus)
DECLARE_PCI_UNMAP_LEN(len)
};
@@ -116,9 +122,14 @@
u8 __iomem *wc_fifo; /* w/c rx dma addr fifo address */
struct mcp_kreq_ether_recv *shadow; /* host shadow of recv ring */
struct myri10ge_rx_buffer_state *info;
+ struct page *page;
+ dma_addr_t bus;
+ int page_offset;
int cnt;
+ int fill_cnt;
int alloc_fail;
int mask; /* number of rx slots -1 */
+ int watchdog_needed;
};
struct myri10ge_tx_buf {
@@ -150,6 +161,7 @@
struct myri10ge_rx_buf rx_big;
struct myri10ge_rx_done rx_done;
int small_bytes;
+ int big_bytes;
struct net_device *dev;
struct net_device_stats stats;
u8 __iomem *sram;
@@ -266,6 +278,10 @@
module_param(myri10ge_debug, int, 0);
MODULE_PARM_DESC(myri10ge_debug, "Debug level (0=none,...,16=all)");
+static int myri10ge_fill_thresh = 256;
+module_param(myri10ge_fill_thresh, int, S_IRUGO | S_IWUSR);
+MODULE_PARM_DESC(myri10ge_fill_thresh, "Number of empty rx slots allowed\n");
+
#define MYRI10GE_FW_OFFSET 1024*1024
#define MYRI10GE_HIGHPART_TO_U32(X) \
(sizeof (X) == 8) ? ((u32)((u64)(X) >> 32)) : (0)
@@ -958,6 +974,180 @@
}
}
+static inline void
+myri10ge_rx_skb_build(struct sk_buff *skb, u8 * va,
+ struct skb_frag_struct *rx_frags, int len, int hlen)
+{
+ struct skb_frag_struct *skb_frags;
+
+ skb->len = skb->data_len = len;
+ skb->truesize = len + sizeof(struct sk_buff);
+ /* attach the page(s) */
+
+ skb_frags = skb_shinfo(skb)->frags;
+ while (len > 0) {
+ memcpy(skb_frags, rx_frags, sizeof(*skb_frags));
+ len -= rx_frags->size;
+ skb_frags++;
+ rx_frags++;
+ skb_shinfo(skb)->nr_frags++;
+ }
+
+ /* pskb_may_pull is not available in irq context, but
+ * skb_pull() (for ether_pad and eth_type_trans()) requires
+ * the beginning of the packet in skb_headlen(), move it
+ * manually */
+ memcpy(skb->data, va, hlen);
+ skb_shinfo(skb)->frags[0].page_offset += hlen;
+ skb_shinfo(skb)->frags[0].size -= hlen;
+ skb->data_len -= hlen;
+ skb->tail += hlen;
+ skb_pull(skb, MXGEFW_PAD);
+}
+
+static void
+myri10ge_alloc_rx_pages(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx,
+ int bytes, int watchdog)
+{
+ struct page *page;
+ int idx;
+
+ if (unlikely(rx->watchdog_needed && !watchdog))
+ return;
+
+ /* try to refill entire ring */
+ while (rx->fill_cnt != (rx->cnt + rx->mask + 1)) {
+ idx = rx->fill_cnt & rx->mask;
+
+ if ((bytes < MYRI10GE_ALLOC_SIZE / 2) &&
+ (rx->page_offset + bytes <= MYRI10GE_ALLOC_SIZE)) {
+ /* we can use part of previous page */
+ get_page(rx->page);
+ } else {
+ /* we need a new page */
+ page =
+ alloc_pages(GFP_ATOMIC | __GFP_COMP,
+ MYRI10GE_ALLOC_ORDER);
+ if (unlikely(page == NULL)) {
+ if (rx->fill_cnt - rx->cnt < 16)
+ rx->watchdog_needed = 1;
+ return;
+ }
+ rx->page = page;
+ rx->page_offset = 0;
+ rx->bus = pci_map_page(mgp->pdev, page, 0,
+ MYRI10GE_ALLOC_SIZE,
+ PCI_DMA_FROMDEVICE);
+ }
+ rx->info[idx].page = rx->page;
+ rx->info[idx].page_offset = rx->page_offset;
+ /* note that this is the address of the start of the
+ * page */
+ pci_unmap_addr_set(&rx->info[idx], bus, rx->bus);
+ rx->shadow[idx].addr_low =
+ htonl(MYRI10GE_LOWPART_TO_U32(rx->bus) + rx->page_offset);
+ rx->shadow[idx].addr_high =
+ htonl(MYRI10GE_HIGHPART_TO_U32(rx->bus));
+
+ /* start next packet on a cacheline boundary */
+ rx->page_offset += SKB_DATA_ALIGN(bytes);
+ rx->fill_cnt++;
+
+ /* copy 8 descriptors to the firmware at a time */
+ if ((idx & 7) == 7) {
+ if (rx->wc_fifo == NULL)
+ myri10ge_submit_8rx(&rx->lanai[idx - 7],
+ &rx->shadow[idx - 7]);
+ else {
+ mb();
+ myri10ge_pio_copy(rx->wc_fifo,
+ &rx->shadow[idx - 7], 64);
+ }
+ }
+ }
+}
+
+static inline void
+myri10ge_unmap_rx_page(struct pci_dev *pdev,
+ struct myri10ge_rx_buffer_state *info, int bytes)
+{
+ /* unmap the recvd page if we're the only or last user of it */
+ if (bytes >= MYRI10GE_ALLOC_SIZE / 2 ||
+ (info->page_offset + 2 * bytes) > MYRI10GE_ALLOC_SIZE) {
+ pci_unmap_page(pdev, (pci_unmap_addr(info, bus)
+ & ~(MYRI10GE_ALLOC_SIZE - 1)),
+ MYRI10GE_ALLOC_SIZE, PCI_DMA_FROMDEVICE);
+ }
+}
+
+#define MYRI10GE_HLEN 64 /* The number of bytes to copy from a
+ * page into an skb */
+
+static inline int
+myri10ge_page_rx_done(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx,
+ int bytes, int len, __wsum csum)
+{
+ struct sk_buff *skb;
+ struct skb_frag_struct rx_frags[MYRI10GE_MAX_FRAGS_PER_FRAME];
+ int i, idx, hlen, remainder;
+ struct pci_dev *pdev = mgp->pdev;
+ struct net_device *dev = mgp->dev;
+ u8 *va;
+
+ len += MXGEFW_PAD;
+ idx = rx->cnt & rx->mask;
+ va = page_address(rx->info[idx].page) + rx->info[idx].page_offset;
+ prefetch(va);
+ /* Fill skb_frag_struct(s) with data from our receive */
+ for (i = 0, remainder = len; remainder > 0; i++) {
+ myri10ge_unmap_rx_page(pdev, &rx->info[idx], bytes);
+ rx_frags[i].page = rx->info[idx].page;
+ rx_frags[i].page_offset = rx->info[idx].page_offset;
+ if (remainder < MYRI10GE_ALLOC_SIZE)
+ rx_frags[i].size = remainder;
+ else
+ rx_frags[i].size = MYRI10GE_ALLOC_SIZE;
+ rx->cnt++;
+ idx = rx->cnt & rx->mask;
+ remainder -= MYRI10GE_ALLOC_SIZE;
+ }
+
+ hlen = MYRI10GE_HLEN > len ? len : MYRI10GE_HLEN;
+
+ /* allocate an skb to attach the page(s) to. */
+
+ skb = netdev_alloc_skb(dev, MYRI10GE_HLEN + 16);
+ if (unlikely(skb == NULL)) {
+ mgp->stats.rx_dropped++;
+ do {
+ i--;
+ put_page(rx_frags[i].page);
+ } while (i != 0);
+ return 0;
+ }
+
+ /* Attach the pages to the skb, and trim off any padding */
+ myri10ge_rx_skb_build(skb, va, rx_frags, len, hlen);
+ if (skb_shinfo(skb)->frags[0].size <= 0) {
+ put_page(skb_shinfo(skb)->frags[0].page);
+ skb_shinfo(skb)->nr_frags = 0;
+ }
+ skb->protocol = eth_type_trans(skb, dev);
+ skb->dev = dev;
+
+ if (mgp->csum_flag) {
+ if ((skb->protocol == htons(ETH_P_IP)) ||
+ (skb->protocol == htons(ETH_P_IPV6))) {
+ skb->csum = csum;
+ skb->ip_summed = CHECKSUM_COMPLETE;
+ } else
+ myri10ge_vlan_ip_csum(skb, csum);
+ }
+ netif_receive_skb(skb);
+ dev->last_rx = jiffies;
+ return 1;
+}
+
static inline unsigned long
myri10ge_rx_done(struct myri10ge_priv *mgp, struct myri10ge_rx_buf *rx,
int bytes, int len, __wsum csum)
next prev parent reply other threads:[~2006-12-11 10:25 UTC|newest]
Thread overview: 9+ messages / expand[flat|nested] mbox.gz Atom feed top
2006-12-11 10:23 [PATCH 0/7] myri10ge updates Brice Goglin
2006-12-11 10:24 ` [PATCH 1/7] myri10ge: indentation cleanups Brice Goglin
2006-12-11 14:54 ` Jeff Garzik
2006-12-11 10:25 ` Brice Goglin [this message]
2006-12-11 10:25 ` [PATCH 3/7] myri10ge: switch to page-based skb Brice Goglin
2006-12-11 10:26 ` [PATCH 4/7] myri10ge: drop contiguous skb routines Brice Goglin
2006-12-11 10:26 ` [PATCH 5/7] myri10ge: Full vlan frame in small_bytes Brice Goglin
2006-12-11 10:27 ` [PATCH 6/7] myri10ge: fix big_bytes in case of vlan frames Brice Goglin
2006-12-11 10:27 ` [PATCH 7/7] myri10ge: update driver version to 1.1.0 Brice Goglin
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=457D3205.6000203@myri.com \
--to=brice@myri.com \
--cc=jeff@garzik.org \
--cc=netdev@vger.kernel.org \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is an external index of several public inboxes,
see mirroring instructions on how to clone and mirror
all data and code used by this external index.