[PATCH net v2 0/1] Allow TX timestamp with UDP GSO

[PATCH net v2 0/1] Allow TX timestamp with UDP GSO

[Fred Klassen]

Fixes an issue where TX Timestamps are not arriving on the error queue
when UDP_SEGMENT CMSG type is combined with CMSG type SO_TIMESTAMPING.

This is version 2 which moves the tests to net-next.

Fred Klassen (1):
  net/udp_gso: Allow TX timestamp with UDP GSO

 net/ipv4/udp_offload.c | 4 ++++
 1 file changed, 4 insertions(+)

-- 
2.11.0

[PATCH net v2 1/1] net/udp_gso: Allow TX timestamp with UDP GSO

[Fred Klassen]

Fixes an issue where TX Timestamps are not arriving on the error queue
when UDP_SEGMENT CMSG type is combined with CMSG type SO_TIMESTAMPING.
This can be illustrated with an updated updgso_bench_tx program which
includes the '-T' option to test for this condition.

    ./udpgso_bench_tx -4ucTPv -S 1472 -l2 -D 172.16.120.18
    poll timeout
    udp tx:      0 MB/s        1 calls/s      1 msg/s

The "poll timeout" message above indicates that TX timestamp never
arrived.

It also appears that other TX CMSG types cause similar issues, for
example trying to set SOL_IP/IP_TOS.

    ./udpgso_bench_tx -4ucPv -S 1472 -q 182 -l2 -D 172.16.120.18
    poll timeout
    udp tx:      0 MB/s        1 calls/s      1 msg/s

This patch preserves tx_flags for the first UDP GSO segment. This
mirrors the stack's behaviour for IPv4 fragments.

Fixes: ee80d1ebe5ba ("udp: add udp gso")
Signed-off-by: Fred Klassen <fklassen@appneta.com>
---
 net/ipv4/udp_offload.c | 4 ++++
 1 file changed, 4 insertions(+)

diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
index 065334b41d57..33de347695ae 100644
--- a/net/ipv4/udp_offload.c
+++ b/net/ipv4/udp_offload.c
@@ -228,6 +228,10 @@ struct sk_buff *__udp_gso_segment(struct sk_buff *gso_skb,
 	seg = segs;
 	uh = udp_hdr(seg);
 
+	/* preserve TX timestamp and zero-copy info for first segment */
+	skb_shinfo(seg)->tskey = skb_shinfo(gso_skb)->tskey;
+	skb_shinfo(seg)->tx_flags = skb_shinfo(gso_skb)->tx_flags;
+
 	/* compute checksum adjustment based on old length versus new */
 	newlen = htons(sizeof(*uh) + mss);
 	check = csum16_add(csum16_sub(uh->check, uh->len), newlen);
-- 
2.11.0

Re: [PATCH net v2 1/1] net/udp_gso: Allow TX timestamp with UDP GSO

[Willem de Bruijn]

On Sun, May 26, 2019 at 10:37 AM Fred Klassen <fklassen@appneta.com> wrote:
>
> Fixes an issue where TX Timestamps are not arriving on the error queue
> when UDP_SEGMENT CMSG type is combined with CMSG type SO_TIMESTAMPING.
> This can be illustrated with an updated updgso_bench_tx program which
> includes the '-T' option to test for this condition.
>
>     ./udpgso_bench_tx -4ucTPv -S 1472 -l2 -D 172.16.120.18
>     poll timeout
>     udp tx:      0 MB/s        1 calls/s      1 msg/s
>
> The "poll timeout" message above indicates that TX timestamp never
> arrived.
>
> It also appears that other TX CMSG types cause similar issues, for
> example trying to set SOL_IP/IP_TOS.

If correct we need to find the root cause. Without that, this is not
very informative. And likely the fix is not related, as that does not
involve tx_flags or tskey.

What exact behavior do you observe: tx timestamp notifications go
missing when enabling IP_TOS?

>
>     ./udpgso_bench_tx -4ucPv -S 1472 -q 182 -l2 -D 172.16.120.18
>     poll timeout
>     udp tx:      0 MB/s        1 calls/s      1 msg/s
>
> This patch preserves tx_flags for the first UDP GSO segment. This
> mirrors the stack's behaviour for IPv4 fragments.

But deviates from the established behavior of other transport protocol TCP.

I think it's a bit premature to resubmit as is while still discussing
the original patch.

>
> Fixes: ee80d1ebe5ba ("udp: add udp gso")
> Signed-off-by: Fred Klassen <fklassen@appneta.com>
> ---
>  net/ipv4/udp_offload.c | 4 ++++
>  1 file changed, 4 insertions(+)
>
> diff --git a/net/ipv4/udp_offload.c b/net/ipv4/udp_offload.c
> index 065334b41d57..33de347695ae 100644
> --- a/net/ipv4/udp_offload.c
> +++ b/net/ipv4/udp_offload.c
> @@ -228,6 +228,10 @@ struct sk_buff *__udp_gso_segment(struct sk_buff *gso_skb,
>         seg = segs;
>         uh = udp_hdr(seg);
>
> +       /* preserve TX timestamp and zero-copy info for first segment */

As pointed out in v1, zerocopy flags are handled by protocol
independent skb_segment. It calls skb_zerocopy_clone.



> +       skb_shinfo(seg)->tskey = skb_shinfo(gso_skb)->tskey;
> +       skb_shinfo(seg)->tx_flags = skb_shinfo(gso_skb)->tx_flags;
> +
>         /* compute checksum adjustment based on old length versus new */
>         newlen = htons(sizeof(*uh) + mss);
>         check = csum16_add(csum16_sub(uh->check, uh->len), newlen);
> --
> 2.11.0
>