* [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2)
@ 2011-10-20 9:01 Ian Campbell
2011-10-20 9:01 ` [PATCH 1/6] mm: add a "struct page_frag" type containing a page, offset and length Ian Campbell
` (6 more replies)
0 siblings, 7 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org; +Cc: linux-scsi@vger.kernel.org, linux-mm@kvack.org
The following series is the second attempt to convert a fifth (and
hopefully final) batch of network drivers to the SKB pages fragment API
introduced in 131ea6675c76.
There are four drivers here (mlx4, cxgb4, cxgb4vf and cxgbi) which used
skb_frag_t as part of their internal datastructures which meant that
they are impacted by changes to that type more than most drivers. To
break this dependency I added a "struct page_frag" (struct page + offset
+ len) and converted them to use it. These conversions are a little less
trivial than most of the preceding ones and I have only been able to
compile test them.
The struct page_frag addition has been acked by Andrew Morton to go
through the net tree. (Andrew, I took you "yes please" as an Acked-by. I
hope that's ok).
The final patch here wraps the page member of skb_frag_t in a structure,
this is a precursor to adding the destructor here (those patches need a
little more work, arising from comments made at LPC, I'll post regarding
those shortly). This should help ensure that no direct uses of the page
get introduced in the meantime.
In the previous posting of this series I ran an allmodconfig build on a
boatload architectures[2] on a baseline of the then current
net-next/master (88c5100c28b0) and with that series. Although the
baseline didn't build for most architectures I used "make -k" and
confirmed that this series added no new warnings or errors. For this
iteration I have just rebuilt things which changed in the interval
88c5100c28b0..a0bec1cd8f7a (current net-next/master) on amd64 and
eyeballed the diff for new uses of frag->page (I saw none).
This is part of my series to enable visibility into SKB paged fragment's
lifecycles, [0] contains some more background and rationale but
basically the completed series will allow entities which inject pages
into the networking stack to receive a notification when the stack has
really finished with those pages (i.e. including retransmissions,
clones, pull-ups etc) and not just when the original skb is finished
with, which is beneficial to many subsystems which wish to inject pages
into the network stack without giving up full ownership of those page's
lifecycle. It implements something broadly along the lines of what was
described in [1].
Cheers,
Ian.
[0] http://marc.info/?l=linux-netdev&m=131072801125521&w=2
[1] http://marc.info/?l=linux-netdev&m=130925719513084&w=2
[2] arm amd64 blackfin cris i386 ia64 m68k mips64 mips powerpc64 powerpc
s390x sh4 sparc64 sparc xtensa
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 8+ messages in thread
* [PATCH 1/6] mm: add a "struct page_frag" type containing a page, offset and length
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:01 ` [PATCH 2/6] mlx4: convert to SKB paged frag API Ian Campbell
` (5 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org
Cc: Ian Campbell, Christoph Hellwig, David Miller, linux-mm,
linux-kernel
A few network drivers currently use skb_frag_struct for this purpose but I have
patches which add additional fields and semantics there which these other uses
do not want.
A structure for reference sub-page regions seems like a generally useful thing
so do so instead of adding a network subsystem specific structure.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Acked-by: Jens Axboe <jaxboe@fusionio.com>
Acked-by: David Rientjes <rientjes@google.com>
Acked-by: Andrew Morton <akpm@linux-foundation.org>
Cc: Christoph Hellwig <hch@infradead.org>
Cc: David Miller <davem@davemloft.net>
Cc: linux-mm@kvack.org
Cc: linux-kernel@vger.kernel.org
[since v1: s/struct subpage/struct page_frag/ on advice from Christoph]
[since v2: s/page_offset/offset/ on advice from Andrew]
---
include/linux/mm_types.h | 11 +++++++++++
1 files changed, 11 insertions(+), 0 deletions(-)
diff --git a/include/linux/mm_types.h b/include/linux/mm_types.h
index 774b895..29971a5 100644
--- a/include/linux/mm_types.h
+++ b/include/linux/mm_types.h
@@ -135,6 +135,17 @@ struct page {
#endif
;
+struct page_frag {
+ struct page *page;
+#if (BITS_PER_LONG > 32) || (PAGE_SIZE >= 65536)
+ __u32 offset;
+ __u32 size;
+#else
+ __u16 offset;
+ __u16 size;
+#endif
+};
+
typedef unsigned long __nocast vm_flags_t;
/*
--
1.7.2.5
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 2/6] mlx4: convert to SKB paged frag API.
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
2011-10-20 9:01 ` [PATCH 1/6] mm: add a "struct page_frag" type containing a page, offset and length Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:01 ` [PATCH 3/6] cxgb4: " Ian Campbell
` (4 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org; +Cc: Ian Campbell, netdev
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: netdev@vger.kernel.org
---
drivers/net/ethernet/mellanox/mlx4/en_rx.c | 32 ++++++++++++++--------------
drivers/net/ethernet/mellanox/mlx4/en_tx.c | 20 +++--------------
2 files changed, 20 insertions(+), 32 deletions(-)
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_rx.c b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
index 9b18d85..af26983 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_rx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_rx.c
@@ -44,7 +44,7 @@
static int mlx4_en_alloc_frag(struct mlx4_en_priv *priv,
struct mlx4_en_rx_desc *rx_desc,
- struct skb_frag_struct *skb_frags,
+ struct page_frag *skb_frags,
struct mlx4_en_rx_alloc *ring_alloc,
int i)
{
@@ -61,7 +61,7 @@ static int mlx4_en_alloc_frag(struct mlx4_en_priv *priv,
return -ENOMEM;
skb_frags[i].page = page_alloc->page;
- skb_frags[i].page_offset = page_alloc->offset;
+ skb_frags[i].offset = page_alloc->offset;
page_alloc->page = page;
page_alloc->offset = frag_info->frag_align;
} else {
@@ -69,11 +69,11 @@ static int mlx4_en_alloc_frag(struct mlx4_en_priv *priv,
get_page(page);
skb_frags[i].page = page;
- skb_frags[i].page_offset = page_alloc->offset;
+ skb_frags[i].offset = page_alloc->offset;
page_alloc->offset += frag_info->frag_stride;
}
dma = pci_map_single(mdev->pdev, page_address(skb_frags[i].page) +
- skb_frags[i].page_offset, frag_info->frag_size,
+ skb_frags[i].offset, frag_info->frag_size,
PCI_DMA_FROMDEVICE);
rx_desc->data[i].addr = cpu_to_be64(dma);
return 0;
@@ -157,8 +157,8 @@ static int mlx4_en_prepare_rx_desc(struct mlx4_en_priv *priv,
struct mlx4_en_rx_ring *ring, int index)
{
struct mlx4_en_rx_desc *rx_desc = ring->buf + (index * ring->stride);
- struct skb_frag_struct *skb_frags = ring->rx_info +
- (index << priv->log_rx_info);
+ struct page_frag *skb_frags = ring->rx_info +
+ (index << priv->log_rx_info);
int i;
for (i = 0; i < priv->num_frags; i++)
@@ -183,7 +183,7 @@ static void mlx4_en_free_rx_desc(struct mlx4_en_priv *priv,
int index)
{
struct mlx4_en_dev *mdev = priv->mdev;
- struct skb_frag_struct *skb_frags;
+ struct page_frag *skb_frags;
struct mlx4_en_rx_desc *rx_desc = ring->buf + (index << ring->log_stride);
dma_addr_t dma;
int nr;
@@ -194,7 +194,7 @@ static void mlx4_en_free_rx_desc(struct mlx4_en_priv *priv,
dma = be64_to_cpu(rx_desc->data[nr].addr);
en_dbg(DRV, priv, "Unmapping buffer at dma:0x%llx\n", (u64) dma);
- pci_unmap_single(mdev->pdev, dma, skb_frag_size(&skb_frags[nr]),
+ pci_unmap_single(mdev->pdev, dma, skb_frags[nr].size,
PCI_DMA_FROMDEVICE);
put_page(skb_frags[nr].page);
}
@@ -403,7 +403,7 @@ void mlx4_en_deactivate_rx_ring(struct mlx4_en_priv *priv,
/* Unmap a completed descriptor and free unused pages */
static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv,
struct mlx4_en_rx_desc *rx_desc,
- struct skb_frag_struct *skb_frags,
+ struct page_frag *skb_frags,
struct skb_frag_struct *skb_frags_rx,
struct mlx4_en_rx_alloc *page_alloc,
int length)
@@ -420,9 +420,9 @@ static int mlx4_en_complete_rx_desc(struct mlx4_en_priv *priv,
break;
/* Save page reference in skb */
- skb_frags_rx[nr].page = skb_frags[nr].page;
- skb_frag_size_set(&skb_frags_rx[nr], skb_frag_size(&skb_frags[nr]));
- skb_frags_rx[nr].page_offset = skb_frags[nr].page_offset;
+ __skb_frag_set_page(&skb_frags_rx[nr], skb_frags[nr].page);
+ skb_frag_size_set(&skb_frags_rx[nr], skb_frags[nr].size);
+ skb_frags_rx[nr].page_offset = skb_frags[nr].offset;
dma = be64_to_cpu(rx_desc->data[nr].addr);
/* Allocate a replacement page */
@@ -444,7 +444,7 @@ fail:
* the descriptor) of this packet; remaining fragments are reused... */
while (nr > 0) {
nr--;
- put_page(skb_frags_rx[nr].page);
+ __skb_frag_unref(&skb_frags_rx[nr]);
}
return 0;
}
@@ -452,7 +452,7 @@ fail:
static struct sk_buff *mlx4_en_rx_skb(struct mlx4_en_priv *priv,
struct mlx4_en_rx_desc *rx_desc,
- struct skb_frag_struct *skb_frags,
+ struct page_frag *skb_frags,
struct mlx4_en_rx_alloc *page_alloc,
unsigned int length)
{
@@ -474,7 +474,7 @@ static struct sk_buff *mlx4_en_rx_skb(struct mlx4_en_priv *priv,
/* Get pointer to first fragment so we could copy the headers into the
* (linear part of the) skb */
- va = page_address(skb_frags[0].page) + skb_frags[0].page_offset;
+ va = page_address(skb_frags[0].page) + skb_frags[0].offset;
if (length <= SMALL_PACKET_SIZE) {
/* We are copying all relevant data to the skb - temporarily
@@ -533,7 +533,7 @@ int mlx4_en_process_rx_cq(struct net_device *dev, struct mlx4_en_cq *cq, int bud
struct mlx4_en_priv *priv = netdev_priv(dev);
struct mlx4_cqe *cqe;
struct mlx4_en_rx_ring *ring = &priv->rx_ring[cq->ring];
- struct skb_frag_struct *skb_frags;
+ struct page_frag *skb_frags;
struct mlx4_en_rx_desc *rx_desc;
struct sk_buff *skb;
int index;
diff --git a/drivers/net/ethernet/mellanox/mlx4/en_tx.c b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
index 75dda26..75338eb 100644
--- a/drivers/net/ethernet/mellanox/mlx4/en_tx.c
+++ b/drivers/net/ethernet/mellanox/mlx4/en_tx.c
@@ -460,26 +460,13 @@ static inline void mlx4_en_xmit_poll(struct mlx4_en_priv *priv, int tx_ind)
}
}
-static void *get_frag_ptr(struct sk_buff *skb)
-{
- struct skb_frag_struct *frag = &skb_shinfo(skb)->frags[0];
- struct page *page = frag->page;
- void *ptr;
-
- ptr = page_address(page);
- if (unlikely(!ptr))
- return NULL;
-
- return ptr + frag->page_offset;
-}
-
static int is_inline(struct sk_buff *skb, void **pfrag)
{
void *ptr;
if (inline_thold && !skb_is_gso(skb) && skb->len <= inline_thold) {
if (skb_shinfo(skb)->nr_frags == 1) {
- ptr = get_frag_ptr(skb);
+ ptr = skb_frag_address_safe(&skb_shinfo(skb)->frags[0]);
if (unlikely(!ptr))
return 0;
@@ -756,8 +743,9 @@ netdev_tx_t mlx4_en_xmit(struct sk_buff *skb, struct net_device *dev)
/* Map fragments */
for (i = skb_shinfo(skb)->nr_frags - 1; i >= 0; i--) {
frag = &skb_shinfo(skb)->frags[i];
- dma = pci_map_page(mdev->dev->pdev, frag->page, frag->page_offset,
- skb_frag_size(frag), PCI_DMA_TODEVICE);
+ dma = skb_frag_dma_map(&mdev->dev->pdev->dev, frag,
+ 0, skb_frag_size(frag),
+ DMA_TO_DEVICE);
data->addr = cpu_to_be64(dma);
data->lkey = cpu_to_be32(mdev->mr.key);
wmb();
--
1.7.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 3/6] cxgb4: convert to SKB paged frag API.
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
2011-10-20 9:01 ` [PATCH 1/6] mm: add a "struct page_frag" type containing a page, offset and length Ian Campbell
2011-10-20 9:01 ` [PATCH 2/6] mlx4: convert to SKB paged frag API Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:01 ` [PATCH 4/6] cxgb4vf: " Ian Campbell
` (3 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org; +Cc: Ian Campbell, Dimitris Michailidis, netdev
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Dimitris Michailidis <dm@chelsio.com>
Cc: netdev@vger.kernel.org
---
drivers/net/ethernet/chelsio/cxgb4/cxgb4.h | 2 +-
drivers/net/ethernet/chelsio/cxgb4/sge.c | 45 ++++++++++++++-------------
2 files changed, 24 insertions(+), 23 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
index 223a7f7..0fe1885 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
+++ b/drivers/net/ethernet/chelsio/cxgb4/cxgb4.h
@@ -326,7 +326,7 @@ struct sge_fl { /* SGE free-buffer queue state */
/* A packet gather list */
struct pkt_gl {
- skb_frag_t frags[MAX_SKB_FRAGS];
+ struct page_frag frags[MAX_SKB_FRAGS];
void *va; /* virtual address of first byte */
unsigned int nfrags; /* # of fragments */
unsigned int tot_len; /* total length of fragments */
diff --git a/drivers/net/ethernet/chelsio/cxgb4/sge.c b/drivers/net/ethernet/chelsio/cxgb4/sge.c
index 14f31d3..ddc1698 100644
--- a/drivers/net/ethernet/chelsio/cxgb4/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4/sge.c
@@ -215,8 +215,8 @@ static int map_skb(struct device *dev, const struct sk_buff *skb,
end = &si->frags[si->nr_frags];
for (fp = si->frags; fp < end; fp++) {
- *++addr = dma_map_page(dev, fp->page, fp->page_offset,
- skb_frag_size(fp), DMA_TO_DEVICE);
+ *++addr = skb_frag_dma_map(dev, fp, 0, skb_frag_size(fp),
+ DMA_TO_DEVICE);
if (dma_mapping_error(dev, *addr))
goto unwind;
}
@@ -1409,22 +1409,23 @@ int cxgb4_ofld_send(struct net_device *dev, struct sk_buff *skb)
}
EXPORT_SYMBOL(cxgb4_ofld_send);
-static inline void copy_frags(struct skb_shared_info *ssi,
+static inline void copy_frags(struct sk_buff *skb,
const struct pkt_gl *gl, unsigned int offset)
{
- unsigned int n;
+ int i;
/* usually there's just one frag */
- ssi->frags[0].page = gl->frags[0].page;
- ssi->frags[0].page_offset = gl->frags[0].page_offset + offset;
- skb_frag_size_set(&ssi->frags[0], skb_frag_size(&gl->frags[0]) - offset);
- ssi->nr_frags = gl->nfrags;
- n = gl->nfrags - 1;
- if (n)
- memcpy(&ssi->frags[1], &gl->frags[1], n * sizeof(skb_frag_t));
+ __skb_fill_page_desc(skb, 0, gl->frags[0].page,
+ gl->frags[0].offset + offset,
+ gl->frags[0].size - offset);
+ skb_shinfo(skb)->nr_frags = gl->nfrags;
+ for (i = 1; i < gl->nfrags; i++)
+ __skb_fill_page_desc(skb, i, gl->frags[i].page,
+ gl->frags[i].offset,
+ gl->frags[i].size);
/* get a reference to the last page, we don't own it */
- get_page(gl->frags[n].page);
+ get_page(gl->frags[gl->nfrags - 1].page);
}
/**
@@ -1459,7 +1460,7 @@ struct sk_buff *cxgb4_pktgl_to_skb(const struct pkt_gl *gl,
__skb_put(skb, pull_len);
skb_copy_to_linear_data(skb, gl->va, pull_len);
- copy_frags(skb_shinfo(skb), gl, pull_len);
+ copy_frags(skb, gl, pull_len);
skb->len = gl->tot_len;
skb->data_len = skb->len - pull_len;
skb->truesize += skb->data_len;
@@ -1478,7 +1479,7 @@ EXPORT_SYMBOL(cxgb4_pktgl_to_skb);
static void t4_pktgl_free(const struct pkt_gl *gl)
{
int n;
- const skb_frag_t *p;
+ const struct page_frag *p;
for (p = gl->frags, n = gl->nfrags - 1; n--; p++)
put_page(p->page);
@@ -1522,7 +1523,7 @@ static void do_gro(struct sge_eth_rxq *rxq, const struct pkt_gl *gl,
return;
}
- copy_frags(skb_shinfo(skb), gl, RX_PKT_PAD);
+ copy_frags(skb, gl, RX_PKT_PAD);
skb->len = gl->tot_len - RX_PKT_PAD;
skb->data_len = skb->len;
skb->truesize += skb->data_len;
@@ -1698,7 +1699,7 @@ static int process_responses(struct sge_rspq *q, int budget)
rmb();
rsp_type = RSPD_TYPE(rc->type_gen);
if (likely(rsp_type == RSP_TYPE_FLBUF)) {
- skb_frag_t *fp;
+ struct page_frag *fp;
struct pkt_gl si;
const struct rx_sw_desc *rsd;
u32 len = ntohl(rc->pldbuflen_qid), bufsz, frags;
@@ -1717,9 +1718,9 @@ static int process_responses(struct sge_rspq *q, int budget)
rsd = &rxq->fl.sdesc[rxq->fl.cidx];
bufsz = get_buf_size(rsd);
fp->page = rsd->page;
- fp->page_offset = q->offset;
- skb_frag_size_set(fp, min(bufsz, len));
- len -= skb_frag_size(fp);
+ fp->offset = q->offset;
+ fp->size = min(bufsz, len);
+ len -= fp->size;
if (!len)
break;
unmap_rx_buf(q->adap, &rxq->fl);
@@ -1731,16 +1732,16 @@ static int process_responses(struct sge_rspq *q, int budget)
*/
dma_sync_single_for_cpu(q->adap->pdev_dev,
get_buf_addr(rsd),
- skb_frag_size(fp), DMA_FROM_DEVICE);
+ fp->size, DMA_FROM_DEVICE);
si.va = page_address(si.frags[0].page) +
- si.frags[0].page_offset;
+ si.frags[0].offset;
prefetch(si.va);
si.nfrags = frags + 1;
ret = q->handler(q, q->cur_desc, &si);
if (likely(ret == 0))
- q->offset += ALIGN(skb_frag_size(fp), FL_ALIGN);
+ q->offset += ALIGN(fp->size, FL_ALIGN);
else
restore_rx_bufs(&si, &rxq->fl, frags);
} else if (likely(rsp_type == RSP_TYPE_CPL)) {
--
1.7.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 4/6] cxgb4vf: convert to SKB paged frag API.
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
` (2 preceding siblings ...)
2011-10-20 9:01 ` [PATCH 3/6] cxgb4: " Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:01 ` [PATCH 5/6] cxgbi: " Ian Campbell
` (2 subsequent siblings)
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org; +Cc: Ian Campbell, Casey Leedom, netdev
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: Casey Leedom <leedom@chelsio.com>
Cc: netdev@vger.kernel.org
---
drivers/net/ethernet/chelsio/cxgb4vf/adapter.h | 2 +-
drivers/net/ethernet/chelsio/cxgb4vf/sge.c | 92 ++++++++++-------------
2 files changed, 41 insertions(+), 53 deletions(-)
diff --git a/drivers/net/ethernet/chelsio/cxgb4vf/adapter.h b/drivers/net/ethernet/chelsio/cxgb4vf/adapter.h
index 594334d..611396c 100644
--- a/drivers/net/ethernet/chelsio/cxgb4vf/adapter.h
+++ b/drivers/net/ethernet/chelsio/cxgb4vf/adapter.h
@@ -144,7 +144,7 @@ struct sge_fl {
* An ingress packet gather list.
*/
struct pkt_gl {
- skb_frag_t frags[MAX_SKB_FRAGS];
+ struct page_frag frags[MAX_SKB_FRAGS];
void *va; /* virtual address of first byte */
unsigned int nfrags; /* # of fragments */
unsigned int tot_len; /* total length of fragments */
diff --git a/drivers/net/ethernet/chelsio/cxgb4vf/sge.c b/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
index c2d456d..8d5d55a 100644
--- a/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
+++ b/drivers/net/ethernet/chelsio/cxgb4vf/sge.c
@@ -296,8 +296,8 @@ static int map_skb(struct device *dev, const struct sk_buff *skb,
si = skb_shinfo(skb);
end = &si->frags[si->nr_frags];
for (fp = si->frags; fp < end; fp++) {
- *++addr = dma_map_page(dev, fp->page, fp->page_offset,
- skb_frag_size(fp), DMA_TO_DEVICE);
+ *++addr = skb_frag_dma_map(dev, fp, 0, skb_frag_size(fp),
+ DMA_TO_DEVICE);
if (dma_mapping_error(dev, *addr))
goto unwind;
}
@@ -1357,6 +1357,35 @@ out_free:
}
/**
+ * copy_frags - copy fragments from gather list into skb_shared_info
+ * @skb: destination skb
+ * @gl: source internal packet gather list
+ * @offset: packet start offset in first page
+ *
+ * Copy an internal packet gather list into a Linux skb_shared_info
+ * structure.
+ */
+static inline void copy_frags(struct sk_buff *skb,
+ const struct pkt_gl *gl,
+ unsigned int offset)
+{
+ int i;
+
+ /* usually there's just one frag */
+ __skb_fill_page_desc(skb, 0, gl->frags[0].page,
+ gl->frags[0].offset + offset,
+ gl->frags[0].size - offset);
+ skb_shinfo(skb)->nr_frags = gl->nfrags;
+ for (i = 1; i < gl->nfrags; i++)
+ __skb_fill_page_desc(skb, i, gl->frags[i].page,
+ gl->frags[i].offset,
+ gl->frags[i].size);
+
+ /* get a reference to the last page, we don't own it */
+ get_page(gl->frags[gl->nfrags - 1].page);
+}
+
+/**
* t4vf_pktgl_to_skb - build an sk_buff from a packet gather list
* @gl: the gather list
* @skb_len: size of sk_buff main body if it carries fragments
@@ -1369,7 +1398,6 @@ struct sk_buff *t4vf_pktgl_to_skb(const struct pkt_gl *gl,
unsigned int skb_len, unsigned int pull_len)
{
struct sk_buff *skb;
- struct skb_shared_info *ssi;
/*
* If the ingress packet is small enough, allocate an skb large enough
@@ -1396,21 +1424,10 @@ struct sk_buff *t4vf_pktgl_to_skb(const struct pkt_gl *gl,
__skb_put(skb, pull_len);
skb_copy_to_linear_data(skb, gl->va, pull_len);
- ssi = skb_shinfo(skb);
- ssi->frags[0].page = gl->frags[0].page;
- ssi->frags[0].page_offset = gl->frags[0].page_offset + pull_len;
- skb_frag_size_set(&ssi->frags[0], skb_frag_size(&gl->frags[0]) - pull_len);
- if (gl->nfrags > 1)
- memcpy(&ssi->frags[1], &gl->frags[1],
- (gl->nfrags-1) * sizeof(skb_frag_t));
- ssi->nr_frags = gl->nfrags;
-
+ copy_frags(skb, gl, pull_len);
skb->len = gl->tot_len;
skb->data_len = skb->len - pull_len;
skb->truesize += skb->data_len;
-
- /* Get a reference for the last page, we don't own it */
- get_page(gl->frags[gl->nfrags - 1].page);
}
out:
@@ -1434,35 +1451,6 @@ void t4vf_pktgl_free(const struct pkt_gl *gl)
}
/**
- * copy_frags - copy fragments from gather list into skb_shared_info
- * @si: destination skb shared info structure
- * @gl: source internal packet gather list
- * @offset: packet start offset in first page
- *
- * Copy an internal packet gather list into a Linux skb_shared_info
- * structure.
- */
-static inline void copy_frags(struct skb_shared_info *si,
- const struct pkt_gl *gl,
- unsigned int offset)
-{
- unsigned int n;
-
- /* usually there's just one frag */
- si->frags[0].page = gl->frags[0].page;
- si->frags[0].page_offset = gl->frags[0].page_offset + offset;
- skb_frag_size_set(&si->frags[0], skb_frag_size(&gl->frags[0]) - offset);
- si->nr_frags = gl->nfrags;
-
- n = gl->nfrags - 1;
- if (n)
- memcpy(&si->frags[1], &gl->frags[1], n * sizeof(skb_frag_t));
-
- /* get a reference to the last page, we don't own it */
- get_page(gl->frags[n].page);
-}
-
-/**
* do_gro - perform Generic Receive Offload ingress packet processing
* @rxq: ingress RX Ethernet Queue
* @gl: gather list for ingress packet
@@ -1484,7 +1472,7 @@ static void do_gro(struct sge_eth_rxq *rxq, const struct pkt_gl *gl,
return;
}
- copy_frags(skb_shinfo(skb), gl, PKTSHIFT);
+ copy_frags(skb, gl, PKTSHIFT);
skb->len = gl->tot_len - PKTSHIFT;
skb->data_len = skb->len;
skb->truesize += skb->data_len;
@@ -1667,7 +1655,7 @@ int process_responses(struct sge_rspq *rspq, int budget)
rmb();
rsp_type = RSPD_TYPE(rc->type_gen);
if (likely(rsp_type == RSP_TYPE_FLBUF)) {
- skb_frag_t *fp;
+ struct page_frag *fp;
struct pkt_gl gl;
const struct rx_sw_desc *sdesc;
u32 bufsz, frag;
@@ -1701,9 +1689,9 @@ int process_responses(struct sge_rspq *rspq, int budget)
sdesc = &rxq->fl.sdesc[rxq->fl.cidx];
bufsz = get_buf_size(sdesc);
fp->page = sdesc->page;
- fp->page_offset = rspq->offset;
- skb_frag_size_set(fp, min(bufsz, len));
- len -= skb_frag_size(fp);
+ fp->offset = rspq->offset;
+ fp->size = min(bufsz, len);
+ len -= fp->size;
if (!len)
break;
unmap_rx_buf(rspq->adapter, &rxq->fl);
@@ -1717,9 +1705,9 @@ int process_responses(struct sge_rspq *rspq, int budget)
*/
dma_sync_single_for_cpu(rspq->adapter->pdev_dev,
get_buf_addr(sdesc),
- skb_frag_size(fp), DMA_FROM_DEVICE);
+ fp->size, DMA_FROM_DEVICE);
gl.va = (page_address(gl.frags[0].page) +
- gl.frags[0].page_offset);
+ gl.frags[0].offset);
prefetch(gl.va);
/*
@@ -1728,7 +1716,7 @@ int process_responses(struct sge_rspq *rspq, int budget)
*/
ret = rspq->handler(rspq, rspq->cur_desc, &gl);
if (likely(ret == 0))
- rspq->offset += ALIGN(skb_frag_size(fp), FL_ALIGN);
+ rspq->offset += ALIGN(fp->size, FL_ALIGN);
else
restore_rx_bufs(&gl, &rxq->fl, frag);
} else if (likely(rsp_type == RSP_TYPE_CPL)) {
--
1.7.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 5/6] cxgbi: convert to SKB paged frag API.
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
` (3 preceding siblings ...)
2011-10-20 9:01 ` [PATCH 4/6] cxgb4vf: " Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:01 ` [PATCH 6/6] net: add opaque struct around skb frag page Ian Campbell
2011-10-20 9:23 ` [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) David Miller
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org
Cc: Ian Campbell, James E.J. Bottomley, David S. Miller,
Mike Christie, James Bottomley, Karen Xie, linux-scsi, netdev
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
Cc: "James E.J. Bottomley" <JBottomley@parallels.com>
Cc: "David S. Miller" <davem@davemloft.net>
Cc: Mike Christie <michaelc@cs.wisc.edu>
Cc: James Bottomley <James.Bottomley@suse.de>
Cc: Karen Xie <kxie@chelsio.com>
Cc: linux-scsi@vger.kernel.org
Cc: netdev@vger.kernel.org
---
drivers/scsi/cxgbi/libcxgbi.c | 28 +++++++++++++++-------------
drivers/scsi/cxgbi/libcxgbi.h | 2 +-
2 files changed, 16 insertions(+), 14 deletions(-)
diff --git a/drivers/scsi/cxgbi/libcxgbi.c b/drivers/scsi/cxgbi/libcxgbi.c
index be69da3..1c1329b 100644
--- a/drivers/scsi/cxgbi/libcxgbi.c
+++ b/drivers/scsi/cxgbi/libcxgbi.c
@@ -1787,7 +1787,7 @@ static int sgl_seek_offset(struct scatterlist *sgl, unsigned int sgcnt,
}
static int sgl_read_to_frags(struct scatterlist *sg, unsigned int sgoffset,
- unsigned int dlen, skb_frag_t *frags,
+ unsigned int dlen, struct page_frag *frags,
int frag_max)
{
unsigned int datalen = dlen;
@@ -1814,8 +1814,8 @@ static int sgl_read_to_frags(struct scatterlist *sg, unsigned int sgoffset,
copy = min(datalen, sglen);
if (i && page == frags[i - 1].page &&
sgoffset + sg->offset ==
- frags[i - 1].page_offset + skb_frag_size(&frags[i - 1])) {
- skb_frag_size_add(&frags[i - 1], copy);
+ frags[i - 1].offset + frags[i - 1].size) {
+ frags[i - 1].size += copy;
} else {
if (i >= frag_max) {
pr_warn("too many pages %u, dlen %u.\n",
@@ -1824,8 +1824,8 @@ static int sgl_read_to_frags(struct scatterlist *sg, unsigned int sgoffset,
}
frags[i].page = page;
- frags[i].page_offset = sg->offset + sgoffset;
- skb_frag_size_set(&frags[i], copy);
+ frags[i].offset = sg->offset + sgoffset;
+ frags[i].size = copy;
i++;
}
datalen -= copy;
@@ -1944,15 +1944,15 @@ int cxgbi_conn_init_pdu(struct iscsi_task *task, unsigned int offset,
if (tdata->nr_frags > MAX_SKB_FRAGS ||
(padlen && tdata->nr_frags == MAX_SKB_FRAGS)) {
char *dst = skb->data + task->hdr_len;
- skb_frag_t *frag = tdata->frags;
+ struct page_frag *frag = tdata->frags;
/* data fits in the skb's headroom */
for (i = 0; i < tdata->nr_frags; i++, frag++) {
char *src = kmap_atomic(frag->page,
KM_SOFTIRQ0);
- memcpy(dst, src+frag->page_offset, skb_frag_size(frag));
- dst += skb_frag_size(frag);
+ memcpy(dst, src+frag->offset, frag->size);
+ dst += frag->size;
kunmap_atomic(src, KM_SOFTIRQ0);
}
if (padlen) {
@@ -1962,11 +1962,13 @@ int cxgbi_conn_init_pdu(struct iscsi_task *task, unsigned int offset,
skb_put(skb, count + padlen);
} else {
/* data fit into frag_list */
- for (i = 0; i < tdata->nr_frags; i++)
- get_page(tdata->frags[i].page);
-
- memcpy(skb_shinfo(skb)->frags, tdata->frags,
- sizeof(skb_frag_t) * tdata->nr_frags);
+ for (i = 0; i < tdata->nr_frags; i++) {
+ __skb_fill_page_desc(skb, i,
+ tdata->frags[i].page,
+ tdata->frags[i].offset,
+ tdata->frags[i].size);
+ skb_frag_ref(skb, i);
+ }
skb_shinfo(skb)->nr_frags = tdata->nr_frags;
skb->len += count;
skb->data_len += count;
diff --git a/drivers/scsi/cxgbi/libcxgbi.h b/drivers/scsi/cxgbi/libcxgbi.h
index 9267844..3a25b11 100644
--- a/drivers/scsi/cxgbi/libcxgbi.h
+++ b/drivers/scsi/cxgbi/libcxgbi.h
@@ -574,7 +574,7 @@ struct cxgbi_endpoint {
#define MAX_PDU_FRAGS ((ULP2_MAX_PDU_PAYLOAD + 512 - 1) / 512)
struct cxgbi_task_data {
unsigned short nr_frags;
- skb_frag_t frags[MAX_PDU_FRAGS];
+ struct page_frag frags[MAX_PDU_FRAGS];
struct sk_buff *skb;
unsigned int offset;
unsigned int count;
--
1.7.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* [PATCH 6/6] net: add opaque struct around skb frag page
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
` (4 preceding siblings ...)
2011-10-20 9:01 ` [PATCH 5/6] cxgbi: " Ian Campbell
@ 2011-10-20 9:01 ` Ian Campbell
2011-10-20 9:23 ` [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) David Miller
6 siblings, 0 replies; 8+ messages in thread
From: Ian Campbell @ 2011-10-20 9:01 UTC (permalink / raw)
To: netdev@vger.kernel.org; +Cc: Ian Campbell
I've split this bit out of the skb frag destructor patch since it helps enforce
the use of the fragment API.
Signed-off-by: Ian Campbell <ian.campbell@citrix.com>
---
include/linux/skbuff.h | 10 ++++++----
net/core/skbuff.c | 6 +++---
2 files changed, 9 insertions(+), 7 deletions(-)
diff --git a/include/linux/skbuff.h b/include/linux/skbuff.h
index 1ebf1ea..aec73c1 100644
--- a/include/linux/skbuff.h
+++ b/include/linux/skbuff.h
@@ -140,7 +140,9 @@ struct sk_buff;
typedef struct skb_frag_struct skb_frag_t;
struct skb_frag_struct {
- struct page *page;
+ struct {
+ struct page *p;
+ } page;
#if (BITS_PER_LONG > 32) || (PAGE_SIZE >= 65536)
__u32 page_offset;
__u32 size;
@@ -1175,7 +1177,7 @@ static inline void __skb_fill_page_desc(struct sk_buff *skb, int i,
{
skb_frag_t *frag = &skb_shinfo(skb)->frags[i];
- frag->page = page;
+ frag->page.p = page;
frag->page_offset = off;
skb_frag_size_set(frag, size);
}
@@ -1699,7 +1701,7 @@ static inline void netdev_free_page(struct net_device *dev, struct page *page)
*/
static inline struct page *skb_frag_page(const skb_frag_t *frag)
{
- return frag->page;
+ return frag->page.p;
}
/**
@@ -1785,7 +1787,7 @@ static inline void *skb_frag_address_safe(const skb_frag_t *frag)
*/
static inline void __skb_frag_set_page(skb_frag_t *frag, struct page *page)
{
- frag->page = page;
+ frag->page.p = page;
}
/**
diff --git a/net/core/skbuff.c b/net/core/skbuff.c
index e271040..ca4db40 100644
--- a/net/core/skbuff.c
+++ b/net/core/skbuff.c
@@ -668,14 +668,14 @@ int skb_copy_ubufs(struct sk_buff *skb, gfp_t gfp_mask)
/* skb frags release userspace buffers */
for (i = 0; i < skb_shinfo(skb)->nr_frags; i++)
- put_page(skb_shinfo(skb)->frags[i].page);
+ skb_frag_unref(skb, i);
uarg->callback(uarg);
/* skb frags point to kernel buffers */
for (i = skb_shinfo(skb)->nr_frags; i > 0; i--) {
- skb_shinfo(skb)->frags[i - 1].page_offset = 0;
- skb_shinfo(skb)->frags[i - 1].page = head;
+ __skb_fill_page_desc(skb, i-1, head, 0,
+ skb_shinfo(skb)->frags[i - 1].size);
head = (struct page *)head->private;
}
--
1.7.2.5
^ permalink raw reply related [flat|nested] 8+ messages in thread
* Re: [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2)
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
` (5 preceding siblings ...)
2011-10-20 9:01 ` [PATCH 6/6] net: add opaque struct around skb frag page Ian Campbell
@ 2011-10-20 9:23 ` David Miller
6 siblings, 0 replies; 8+ messages in thread
From: David Miller @ 2011-10-20 9:23 UTC (permalink / raw)
To: Ian.Campbell; +Cc: netdev, linux-scsi, linux-mm
From: Ian Campbell <Ian.Campbell@citrix.com>
Date: Thu, 20 Oct 2011 10:01:15 +0100
> The following series is the second attempt to convert a fifth (and
> hopefully final) batch of network drivers to the SKB pages fragment API
> introduced in 131ea6675c76.
Applied, thanks Ian.
--
To unsubscribe, send a message with 'unsubscribe linux-mm' in
the body to majordomo@kvack.org. For more info on Linux MM,
see: http://www.linux-mm.org/ .
Fight unfair telecom internet charges in Canada: sign http://stopthemeter.ca/
Don't email: <a href=mailto:"dont@kvack.org"> email@kvack.org </a>
^ permalink raw reply [flat|nested] 8+ messages in thread
end of thread, other threads:[~2011-10-20 9:23 UTC | newest]
Thread overview: 8+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2011-10-20 9:01 [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) Ian Campbell
2011-10-20 9:01 ` [PATCH 1/6] mm: add a "struct page_frag" type containing a page, offset and length Ian Campbell
2011-10-20 9:01 ` [PATCH 2/6] mlx4: convert to SKB paged frag API Ian Campbell
2011-10-20 9:01 ` [PATCH 3/6] cxgb4: " Ian Campbell
2011-10-20 9:01 ` [PATCH 4/6] cxgb4vf: " Ian Campbell
2011-10-20 9:01 ` [PATCH 5/6] cxgbi: " Ian Campbell
2011-10-20 9:01 ` [PATCH 6/6] net: add opaque struct around skb frag page Ian Campbell
2011-10-20 9:23 ` [PATCH 0/6] skb fragment API: convert network drivers (part V, take 2) David Miller
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).