Make TPACKET_V3 signal poll when block is closed rather than for every
packet. Side effect is that poll will be signaled when block retire
timer expires which didn't previously happen. Issue was visible when
sending packets at a very low frequency such that all blocks are retired
before packets are received by TPACKET_V3. This caused avoidable packet
loss. The fix ensures that the signal is sent when blocks are closed
which covers the normal path where the block is filled as well as the
path where the timer expires. The case where a block is filled without
moving to the next block (ie. all blocks are full) will still cause poll
to be signaled.
Signed-off-by: Dan Collins <[email protected]>
---
net/packet/af_packet.c | 11 +++++++----
1 file changed, 7 insertions(+), 4 deletions(-)
diff --git a/net/packet/af_packet.c b/net/packet/af_packet.c
index e52a447..6880f34 100644
--- a/net/packet/af_packet.c
+++ b/net/packet/af_packet.c
@@ -785,6 +785,7 @@ static void prb_close_block(struct tpacket_kbdq_core *pkc1,
struct tpacket3_hdr *last_pkt;
struct tpacket_hdr_v1 *h1 = &pbd1->hdr.bh1;
+ struct sock *sk = &po->sk;
if (po->stats.stats3.tp_drops)
status |= TP_STATUS_LOSING;
@@ -809,6 +810,8 @@ static void prb_close_block(struct tpacket_kbdq_core *pkc1,
/* Flush the block */
prb_flush_block(pkc1, pbd1, status);
+ sk->sk_data_ready(sk);
+
pkc1->kactive_blk_num = GET_NEXT_PRB_BLK_NUM(pkc1);
}
@@ -2052,12 +2055,12 @@ static int tpacket_rcv(struct sk_buff *skb, struct net_device *dev,
smp_wmb();
#endif
- if (po->tp_version <= TPACKET_V2)
+ if (po->tp_version <= TPACKET_V2) {
__packet_set_status(po, h.raw, status);
- else
+ sk->sk_data_ready(sk);
+ } else {
prb_clear_blk_fill_status(&po->rx_ring);
-
- sk->sk_data_ready(sk);
+ }
drop_n_restore:
if (skb_head != skb->data && skb_shared(skb)) {
--
2.1.3
From: Dan Collins <[email protected]>
Date: Fri, 19 Dec 2014 16:49:25 +1300
> Make TPACKET_V3 signal poll when block is closed rather than for every
> packet. Side effect is that poll will be signaled when block retire
> timer expires which didn't previously happen. Issue was visible when
> sending packets at a very low frequency such that all blocks are retired
> before packets are received by TPACKET_V3. This caused avoidable packet
> loss. The fix ensures that the signal is sent when blocks are closed
> which covers the normal path where the block is filled as well as the
> path where the timer expires. The case where a block is filled without
> moving to the next block (ie. all blocks are full) will still cause poll
> to be signaled.
>
> Signed-off-by: Dan Collins <[email protected]>
Applied, thanks.
On Dec 22, 2014, at 12:41 PM, David Miller <[email protected]> wrote:
> From: Dan Collins <[email protected]>
> Date: Fri, 19 Dec 2014 16:49:25 +1300
>
>> Make TPACKET_V3 signal poll when block is closed rather than for every
>> packet. Side effect is that poll will be signaled when block retire
>> timer expires which didn't previously happen. Issue was visible when
>> sending packets at a very low frequency such that all blocks are retired
>> before packets are received by TPACKET_V3. This caused avoidable packet
>> loss. The fix ensures that the signal is sent when blocks are closed
>> which covers the normal path where the block is filled as well as the
>> path where the timer expires. The case where a block is filled without
>> moving to the next block (ie. all blocks are full) will still cause poll
>> to be signaled.
>>
>> Signed-off-by: Dan Collins <[email protected]>
>
> Applied, thanks.
Should this be applied to any of the stable kernel branches? The old behavior is the source of some libpcap/tcpdump/etc. complaints; see, for example, the thread that included
http://marc.info/?l=linux-netdev&m=140713966510035&w=2