[PATCH -mm] Documentation: add networking driver's mapping error handling to DMA-API-HOWTO

[PATCH -mm] Documentation: add networking driver's mapping error handling to DMA-API-HOWTO

[FUJITA Tomonori]

I think that it's a good idea to add what drivers should to do exactly
in the DMA mapping failure case.

Seems some networking drivers call dev_kfree_skb() and return
NETDEV_TX_OK, however, some return NETDEV_TX_BUSY without freeing the
skb. (and many don't even check the failure, which we need to fix).

What networking drivers are supposed to do in the case?

=
From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Subject: [PATCH -mm] Documentation: add networking driver's mapping error handling to DMA-API-HOWTO

Adds the concrete DMA mapping error handling for Networking drivers on
the transmit path.

Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
---
 Documentation/DMA-API-HOWTO.txt |    5 +++++
 1 files changed, 5 insertions(+), 0 deletions(-)

diff --git a/Documentation/DMA-API-HOWTO.txt b/Documentation/DMA-API-HOWTO.txt
index 52618ab..8c3235c 100644
--- a/Documentation/DMA-API-HOWTO.txt
+++ b/Documentation/DMA-API-HOWTO.txt
@@ -740,6 +740,11 @@ failure can be determined by:
 		 */
 	}
 
+Networking drivers must call dev_kfree_skb to free the socket buffer
+and return NETDEV_TX_OK if the DMA mapping fails on the transmit hook
+(ndo_start_xmit). This means that the socket buffer is just dropped in
+the failure case.
+
 			   Closing
 
 This document, and the API itself, would not be in it's current
-- 
1.6.5

Re: [PATCH -mm] Documentation: add networking driver's mapping error handling to DMA-API-HOWTO

[David Miller]

From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
Date: Sun, 9 May 2010 13:00:49 +0900

> I think that it's a good idea to add what drivers should to do exactly
> in the DMA mapping failure case.
> 
> Seems some networking drivers call dev_kfree_skb() and return
> NETDEV_TX_OK, however, some return NETDEV_TX_BUSY without freeing the
> skb. (and many don't even check the failure, which we need to fix).
> 
> What networking drivers are supposed to do in the case?

The best thing is probably just to drop, pretending the queue
is full and passing the packet back to the stack is just
full of complications and not worth it, so...

> From: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>
> Subject: [PATCH -mm] Documentation: add networking driver's mapping error handling to DMA-API-HOWTO
> 
> Adds the concrete DMA mapping error handling for Networking drivers on
> the transmit path.
> 
> Signed-off-by: FUJITA Tomonori <fujita.tomonori@lab.ntt.co.jp>

Acked-by: David S. Miller <davem@davemloft.net>