The netif_rx() function frees the skb so we can't dereference it to
save the skb->len.
Signed-off-by: Zhang Shurong <[email protected]>
---
drivers/firewire/net.c | 6 ++++--
1 file changed, 4 insertions(+), 2 deletions(-)
diff --git a/drivers/firewire/net.c b/drivers/firewire/net.c
index 538bd677c254..7a4d1a478e33 100644
--- a/drivers/firewire/net.c
+++ b/drivers/firewire/net.c
@@ -479,7 +479,7 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
struct sk_buff *skb, u16 source_node_id,
bool is_broadcast, u16 ether_type)
{
- int status;
+ int status, len;
switch (ether_type) {
case ETH_P_ARP:
@@ -533,13 +533,15 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
}
skb->protocol = protocol;
}
+
+ len = skb->len;
status = netif_rx(skb);
if (status == NET_RX_DROP) {
net->stats.rx_errors++;
net->stats.rx_dropped++;
} else {
net->stats.rx_packets++;
- net->stats.rx_bytes += skb->len;
+ net->stats.rx_bytes += len;
}
return 0;
--
2.41.0
Hi,
Thanks for the fix.
On Fri, Jun 23, 2023 at 01:39:35PM +0800, Zhang Shurong wrote:
> The netif_rx() function frees the skb so we can't dereference it to
> save the skb->len.
>
> Signed-off-by: Zhang Shurong <[email protected]>
> ---
> drivers/firewire/net.c | 6 ++++--
> 1 file changed, 4 insertions(+), 2 deletions(-)
>
> diff --git a/drivers/firewire/net.c b/drivers/firewire/net.c
> index 538bd677c254..7a4d1a478e33 100644
> --- a/drivers/firewire/net.c
> +++ b/drivers/firewire/net.c
> @@ -479,7 +479,7 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
> struct sk_buff *skb, u16 source_node_id,
> bool is_broadcast, u16 ether_type)
> {
> - int status;
> + int status, len;
>
> switch (ether_type) {
> case ETH_P_ARP:
> @@ -533,13 +533,15 @@ static int fwnet_finish_incoming_packet(struct net_device *net,
> }
> skb->protocol = protocol;
> }
> +
> + len = skb->len;
> status = netif_rx(skb);
> if (status == NET_RX_DROP) {
> net->stats.rx_errors++;
> net->stats.rx_dropped++;
> } else {
> net->stats.rx_packets++;
> - net->stats.rx_bytes += skb->len;
> + net->stats.rx_bytes += len;
> }
>
> return 0;
I'm not good at network subsystem, but as long as reading network core,
I think it better not to access members of sk_buff structure after the
call of netif_rx() since the ownership is already delegated from
firewire-net to the network core to dispatch network protocol. The patch
looks good to me.
Reviewed-by: Takashi Sakamoto <[email protected]>
The most of code in firewire-net comes from old driver
(drivers/ieee1394/eth1394.c), thus the most of linux kernel which
supports ieee1394/firewire subsystem affects the issue, however the
potential issue has never appeared since the skb is released enough later
in the different context. I applied the patch to linux-next branch for
v6.5-rc1.
Thanks
Takashi Sakamoto