[PATCH net] bnx2x: fix DMA unmapping of TSO split BDs

[PATCH net] bnx2x: fix DMA unmapping of TSO split BDs

[Michal Schmidt]

bnx2x triggers warnings with CONFIG_DMA_API_DEBUG=y:

  WARNING: CPU: 0 PID: 2253 at lib/dma-debug.c:887 check_unmap+0xf8/0x920()
  bnx2x 0000:28:00.0: DMA-API: device driver frees DMA memory with
  different size [device address=0x00000000da2b389e] [map size=1490 bytes]
  [unmap size=66 bytes]

The reason is that bnx2x splits a TSO BD into two BDs (headers + data)
using one DMA mapping for both, but it uses only the length of the first
BD when unmapping.

This patch fixes the bug by unmapping the whole length of the two BDs.

Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
---
 drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c | 13 +++++++++----
 1 file changed, 9 insertions(+), 4 deletions(-)

diff --git a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
index ec96130..6b2704f 100644
--- a/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
+++ b/drivers/net/ethernet/broadcom/bnx2x/bnx2x_cmn.c
@@ -160,6 +160,7 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata,
 	struct sk_buff *skb = tx_buf->skb;
 	u16 bd_idx = TX_BD(tx_buf->first_bd), new_cons;
 	int nbd;
+	u16 split_bd_len = 0;
 
 	/* prefetch skb end pointer to speedup dev_kfree_skb() */
 	prefetch(&skb->end);
@@ -167,10 +168,7 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata,
 	DP(NETIF_MSG_TX_DONE, "fp[%d]: pkt_idx %d  buff @(%p)->skb %p\n",
 	   txdata->txq_index, idx, tx_buf, skb);
 
-	/* unmap first bd */
 	tx_start_bd = &txdata->tx_desc_ring[bd_idx].start_bd;
-	dma_unmap_single(&bp->pdev->dev, BD_UNMAP_ADDR(tx_start_bd),
-			 BD_UNMAP_LEN(tx_start_bd), DMA_TO_DEVICE);
 
 	nbd = le16_to_cpu(tx_start_bd->nbd) - 1;
 #ifdef BNX2X_STOP_ON_ERROR
@@ -188,12 +186,19 @@ static u16 bnx2x_free_tx_pkt(struct bnx2x *bp, struct bnx2x_fp_txdata *txdata,
 	--nbd;
 	bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));
 
-	/* ...and the TSO split header bd since they have no mapping */
+	/* TSO headers+data bds share a common mapping. See bnx2x_tx_split() */
 	if (tx_buf->flags & BNX2X_TSO_SPLIT_BD) {
+		tx_data_bd = &txdata->tx_desc_ring[bd_idx].reg_bd;
+		split_bd_len = BD_UNMAP_LEN(tx_data_bd);
 		--nbd;
 		bd_idx = TX_BD(NEXT_TX_IDX(bd_idx));
 	}
 
+	/* unmap first bd */
+	dma_unmap_single(&bp->pdev->dev, BD_UNMAP_ADDR(tx_start_bd),
+			 BD_UNMAP_LEN(tx_start_bd) + split_bd_len,
+			 DMA_TO_DEVICE);
+
 	/* now free frags */
 	while (nbd > 0) {
 
-- 
1.8.4.2

Re: [PATCH net] bnx2x: fix DMA unmapping of TSO split BDs

[Eric Dumazet]

On Thu, 2014-01-09 at 14:36 +0100, Michal Schmidt wrote:
> bnx2x triggers warnings with CONFIG_DMA_API_DEBUG=y:
> 
>   WARNING: CPU: 0 PID: 2253 at lib/dma-debug.c:887 check_unmap+0xf8/0x920()
>   bnx2x 0000:28:00.0: DMA-API: device driver frees DMA memory with
>   different size [device address=0x00000000da2b389e] [map size=1490 bytes]
>   [unmap size=66 bytes]
> 
> The reason is that bnx2x splits a TSO BD into two BDs (headers + data)
> using one DMA mapping for both, but it uses only the length of the first
> BD when unmapping.
> 
> This patch fixes the bug by unmapping the whole length of the two BDs.
> 
> Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
> ---

Reviewed-by: Eric Dumazet <edumazet@google.com>

RE: [PATCH net] bnx2x: fix DMA unmapping of TSO split BDs

[Dmitry Kravkov]

> -----Original Message-----
> From: Eric Dumazet [mailto:eric.dumazet@gmail.com]
> Sent: Thursday, January 09, 2014 7:52 PM
> To: Michal Schmidt
> Cc: netdev@vger.kernel.org; Ariel Elior; Yuval Mintz; Dmitry Kravkov; David
> Miller
> Subject: Re: [PATCH net] bnx2x: fix DMA unmapping of TSO split BDs
> 
> On Thu, 2014-01-09 at 14:36 +0100, Michal Schmidt wrote:
> > bnx2x triggers warnings with CONFIG_DMA_API_DEBUG=y:
> >
> >   WARNING: CPU: 0 PID: 2253 at lib/dma-debug.c:887
> check_unmap+0xf8/0x920()
> >   bnx2x 0000:28:00.0: DMA-API: device driver frees DMA memory with
> >   different size [device address=0x00000000da2b389e] [map size=1490
> bytes]
> >   [unmap size=66 bytes]
> >
> > The reason is that bnx2x splits a TSO BD into two BDs (headers + data)
> > using one DMA mapping for both, but it uses only the length of the
> > first BD when unmapping.
> >
> > This patch fixes the bug by unmapping the whole length of the two BDs.
> >
> > Signed-off-by: Michal Schmidt <mschmidt@redhat.com>
> > ---
> 
> Reviewed-by: Eric Dumazet <edumazet@google.com>
> 

Thanks you, Guys.

Acked-by: Dmitry Kravkov <dmitry@broadcom.com>

Re: [PATCH net] bnx2x: fix DMA unmapping of TSO split BDs

[David Miller]

From: Michal Schmidt <mschmidt@redhat.com>
Date: Thu,  9 Jan 2014 14:36:27 +0100

> bnx2x triggers warnings with CONFIG_DMA_API_DEBUG=y:
> 
>   WARNING: CPU: 0 PID: 2253 at lib/dma-debug.c:887 check_unmap+0xf8/0x920()
>   bnx2x 0000:28:00.0: DMA-API: device driver frees DMA memory with
>   different size [device address=0x00000000da2b389e] [map size=1490 bytes]
>   [unmap size=66 bytes]
> 
> The reason is that bnx2x splits a TSO BD into two BDs (headers + data)
> using one DMA mapping for both, but it uses only the length of the first
> BD when unmapping.
> 
> This patch fixes the bug by unmapping the whole length of the two BDs.
> 
> Signed-off-by: Michal Schmidt <mschmidt@redhat.com>

Applied and queued up for -stable, thanks.