netdev.vger.kernel.org archive mirror
 help / color / mirror / Atom feed
* [PATCH] net: dont use __netdev_alloc_skb for bounce buffer
@ 2012-07-02 18:36 Eric Dumazet
  2012-07-02 20:03 ` Stefan Bader
  2012-07-09  6:52 ` David Miller
  0 siblings, 2 replies; 3+ messages in thread
From: Eric Dumazet @ 2012-07-02 18:36 UTC (permalink / raw)
  To: David Miller; +Cc: netdev, Stefan Bader

From: Eric Dumazet <edumazet@google.com>

commit a1c7fff7e1 (net: netdev_alloc_skb() use build_skb()) broke b44 on
some 64bit machines.

It appears b44 and b43 use __netdev_alloc_skb() instead of alloc_skb()
for their bounce buffers.

There is no need to add an extra NET_SKB_PAD reservation for bounce
buffers :

- In TX path, NET_SKB_PAD is useless

- In RX path in b44, we force a copy of incoming frames if
  GFP_DMA allocations were needed.

Reported-and-bisected-by: Stefan Bader <stefan.bader@canonical.com>
Signed-off-by: Eric Dumazet <edumazet@google.com>
---
 drivers/net/ethernet/broadcom/b44.c  |    4 ++--
 drivers/net/wireless/b43legacy/dma.c |    2 +-
 2 files changed, 3 insertions(+), 3 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/b44.c b/drivers/net/ethernet/broadcom/b44.c
index 46b8b7d..d09c6b5 100644
--- a/drivers/net/ethernet/broadcom/b44.c
+++ b/drivers/net/ethernet/broadcom/b44.c
@@ -656,7 +656,7 @@ static int b44_alloc_rx_skb(struct b44 *bp, int src_idx, u32 dest_idx_unmasked)
 			dma_unmap_single(bp->sdev->dma_dev, mapping,
 					     RX_PKT_BUF_SZ, DMA_FROM_DEVICE);
 		dev_kfree_skb_any(skb);
-		skb = __netdev_alloc_skb(bp->dev, RX_PKT_BUF_SZ, GFP_ATOMIC|GFP_DMA);
+		skb = alloc_skb(RX_PKT_BUF_SZ, GFP_ATOMIC | GFP_DMA);
 		if (skb == NULL)
 			return -ENOMEM;
 		mapping = dma_map_single(bp->sdev->dma_dev, skb->data,
@@ -967,7 +967,7 @@ static netdev_tx_t b44_start_xmit(struct sk_buff *skb, struct net_device *dev)
 			dma_unmap_single(bp->sdev->dma_dev, mapping, len,
 					     DMA_TO_DEVICE);
 
-		bounce_skb = __netdev_alloc_skb(dev, len, GFP_ATOMIC | GFP_DMA);
+		bounce_skb = alloc_skb(len, GFP_ATOMIC | GFP_DMA);
 		if (!bounce_skb)
 			goto err_out;
 
diff --git a/drivers/net/wireless/b43legacy/dma.c b/drivers/net/wireless/b43legacy/dma.c
index f1f8bd0..c8baf02 100644
--- a/drivers/net/wireless/b43legacy/dma.c
+++ b/drivers/net/wireless/b43legacy/dma.c
@@ -1072,7 +1072,7 @@ static int dma_tx_fragment(struct b43legacy_dmaring *ring,
 	meta->dmaaddr = map_descbuffer(ring, skb->data, skb->len, 1);
 	/* create a bounce buffer in zone_dma on mapping failure. */
 	if (b43legacy_dma_mapping_error(ring, meta->dmaaddr, skb->len, 1)) {
-		bounce_skb = __dev_alloc_skb(skb->len, GFP_ATOMIC | GFP_DMA);
+		bounce_skb = alloc_skb(skb->len, GFP_ATOMIC | GFP_DMA);
 		if (!bounce_skb) {
 			ring->current_slot = old_top_slot;
 			ring->used_slots = old_used_slots;

^ permalink raw reply related	[flat|nested] 3+ messages in thread

* Re: [PATCH] net: dont use __netdev_alloc_skb for bounce buffer
  2012-07-02 18:36 [PATCH] net: dont use __netdev_alloc_skb for bounce buffer Eric Dumazet
@ 2012-07-02 20:03 ` Stefan Bader
  2012-07-09  6:52 ` David Miller
  1 sibling, 0 replies; 3+ messages in thread
From: Stefan Bader @ 2012-07-02 20:03 UTC (permalink / raw)
  To: Eric Dumazet; +Cc: David Miller, netdev

[-- Attachment #1: Type: text/plain, Size: 2780 bytes --]

I can confirm that, with the below patch applied, at least the b44 regression is
fixed and network is usable again.

-Stefan

On 02.07.2012 20:36, Eric Dumazet wrote:
> From: Eric Dumazet <edumazet@google.com>
> 
> commit a1c7fff7e1 (net: netdev_alloc_skb() use build_skb()) broke b44 on
> some 64bit machines.
> 
> It appears b44 and b43 use __netdev_alloc_skb() instead of alloc_skb()
> for their bounce buffers.
> 
> There is no need to add an extra NET_SKB_PAD reservation for bounce
> buffers :
> 
> - In TX path, NET_SKB_PAD is useless
> 
> - In RX path in b44, we force a copy of incoming frames if
>   GFP_DMA allocations were needed.
> 
> Reported-and-bisected-by: Stefan Bader <stefan.bader@canonical.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>
> ---
>  drivers/net/ethernet/broadcom/b44.c  |    4 ++--
>  drivers/net/wireless/b43legacy/dma.c |    2 +-
>  2 files changed, 3 insertions(+), 3 deletions(-)
> 
> diff --git a/drivers/net/ethernet/broadcom/b44.c b/drivers/net/ethernet/broadcom/b44.c
> index 46b8b7d..d09c6b5 100644
> --- a/drivers/net/ethernet/broadcom/b44.c
> +++ b/drivers/net/ethernet/broadcom/b44.c
> @@ -656,7 +656,7 @@ static int b44_alloc_rx_skb(struct b44 *bp, int src_idx, u32 dest_idx_unmasked)
>  			dma_unmap_single(bp->sdev->dma_dev, mapping,
>  					     RX_PKT_BUF_SZ, DMA_FROM_DEVICE);
>  		dev_kfree_skb_any(skb);
> -		skb = __netdev_alloc_skb(bp->dev, RX_PKT_BUF_SZ, GFP_ATOMIC|GFP_DMA);
> +		skb = alloc_skb(RX_PKT_BUF_SZ, GFP_ATOMIC | GFP_DMA);
>  		if (skb == NULL)
>  			return -ENOMEM;
>  		mapping = dma_map_single(bp->sdev->dma_dev, skb->data,
> @@ -967,7 +967,7 @@ static netdev_tx_t b44_start_xmit(struct sk_buff *skb, struct net_device *dev)
>  			dma_unmap_single(bp->sdev->dma_dev, mapping, len,
>  					     DMA_TO_DEVICE);
>  
> -		bounce_skb = __netdev_alloc_skb(dev, len, GFP_ATOMIC | GFP_DMA);
> +		bounce_skb = alloc_skb(len, GFP_ATOMIC | GFP_DMA);
>  		if (!bounce_skb)
>  			goto err_out;
>  
> diff --git a/drivers/net/wireless/b43legacy/dma.c b/drivers/net/wireless/b43legacy/dma.c
> index f1f8bd0..c8baf02 100644
> --- a/drivers/net/wireless/b43legacy/dma.c
> +++ b/drivers/net/wireless/b43legacy/dma.c
> @@ -1072,7 +1072,7 @@ static int dma_tx_fragment(struct b43legacy_dmaring *ring,
>  	meta->dmaaddr = map_descbuffer(ring, skb->data, skb->len, 1);
>  	/* create a bounce buffer in zone_dma on mapping failure. */
>  	if (b43legacy_dma_mapping_error(ring, meta->dmaaddr, skb->len, 1)) {
> -		bounce_skb = __dev_alloc_skb(skb->len, GFP_ATOMIC | GFP_DMA);
> +		bounce_skb = alloc_skb(skb->len, GFP_ATOMIC | GFP_DMA);
>  		if (!bounce_skb) {
>  			ring->current_slot = old_top_slot;
>  			ring->used_slots = old_used_slots;
> 
> 




[-- Attachment #2: OpenPGP digital signature --]
[-- Type: application/pgp-signature, Size: 900 bytes --]

^ permalink raw reply	[flat|nested] 3+ messages in thread

* Re: [PATCH] net: dont use __netdev_alloc_skb for bounce buffer
  2012-07-02 18:36 [PATCH] net: dont use __netdev_alloc_skb for bounce buffer Eric Dumazet
  2012-07-02 20:03 ` Stefan Bader
@ 2012-07-09  6:52 ` David Miller
  1 sibling, 0 replies; 3+ messages in thread
From: David Miller @ 2012-07-09  6:52 UTC (permalink / raw)
  To: eric.dumazet; +Cc: netdev, stefan.bader

From: Eric Dumazet <eric.dumazet@gmail.com>
Date: Mon, 02 Jul 2012 20:36:12 +0200

> From: Eric Dumazet <edumazet@google.com>
> 
> commit a1c7fff7e1 (net: netdev_alloc_skb() use build_skb()) broke b44 on
> some 64bit machines.
> 
> It appears b44 and b43 use __netdev_alloc_skb() instead of alloc_skb()
> for their bounce buffers.
> 
> There is no need to add an extra NET_SKB_PAD reservation for bounce
> buffers :
> 
> - In TX path, NET_SKB_PAD is useless
> 
> - In RX path in b44, we force a copy of incoming frames if
>   GFP_DMA allocations were needed.
> 
> Reported-and-bisected-by: Stefan Bader <stefan.bader@canonical.com>
> Signed-off-by: Eric Dumazet <edumazet@google.com>

Applied, thanks Eric.

^ permalink raw reply	[flat|nested] 3+ messages in thread

end of thread, other threads:[~2012-07-09  6:52 UTC | newest]

Thread overview: 3+ messages (download: mbox.gz follow: Atom feed
-- links below jump to the message on this page --
2012-07-02 18:36 [PATCH] net: dont use __netdev_alloc_skb for bounce buffer Eric Dumazet
2012-07-02 20:03 ` Stefan Bader
2012-07-09  6:52 ` 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).