Subject: [PATCH] ath6kl: Fix random rx data corruption

The skb->tail pointer of rx buffers is not adjusted
after skb->data pointer is aligned to 4-byte, this
causes random rx data corruption.

Signed-off-by: Jin Navy <[email protected]>
Signed-off-by: Vasanthakumar Thiagarajan <[email protected]>
---
drivers/net/wireless/ath/ath6kl/txrx.c | 10 ++++++++--
1 files changed, 8 insertions(+), 2 deletions(-)

diff --git a/drivers/net/wireless/ath/ath6kl/txrx.c b/drivers/net/wireless/ath/ath6kl/txrx.c
index efee590..f070876 100644
--- a/drivers/net/wireless/ath/ath6kl/txrx.c
+++ b/drivers/net/wireless/ath/ath6kl/txrx.c
@@ -888,8 +888,11 @@ void ath6kl_rx_refill(struct htc_target *target, enum htc_endpoint_id endpoint)
break;

packet = (struct htc_packet *) skb->head;
- if (!IS_ALIGNED((unsigned long) skb->data, 4))
+ if (!IS_ALIGNED((unsigned long) skb->data, 4)) {
+ size_t len = skb_headlen(skb);
skb->data = PTR_ALIGN(skb->data - 4, 4);
+ skb_set_tail_pointer(skb, len);
+ }
set_htc_rxpkt_info(packet, skb, skb->data,
ATH6KL_BUFFER_SIZE, endpoint);
packet->skb = skb;
@@ -911,8 +914,11 @@ void ath6kl_refill_amsdu_rxbufs(struct ath6kl *ar, int count)
return;

packet = (struct htc_packet *) skb->head;
- if (!IS_ALIGNED((unsigned long) skb->data, 4))
+ if (!IS_ALIGNED((unsigned long) skb->data, 4)) {
+ size_t len = skb_headlen(skb);
skb->data = PTR_ALIGN(skb->data - 4, 4);
+ skb_set_tail_pointer(skb, len);
+ }
set_htc_rxpkt_info(packet, skb, skb->data,
ATH6KL_AMSDU_BUFFER_SIZE, 0);
packet->skb = skb;
--
1.7.0.4



2012-10-12 11:49:07

by Kalle Valo

[permalink] [raw]
Subject: Re: [PATCH] ath6kl: Fix random rx data corruption

On 09/21/2012 10:15 AM, Vasanthakumar Thiagarajan wrote:
> The skb->tail pointer of rx buffers is not adjusted
> after skb->data pointer is aligned to 4-byte, this
> causes random rx data corruption.
>
> Signed-off-by: Jin Navy <[email protected]>
> Signed-off-by: Vasanthakumar Thiagarajan <[email protected]>

Thanks, I have applied this.

This alignment issue has created nothing but grief in ath6kl, at some
point we should really fix this properly. For example, the problem
should only happen in TX path, not in RX path.

Kalle