On Mon, Sep 30, 2019 at 7:57 PM Alexander Duyck
<alexander.du...@gmail.com> wrote:
>
> On Mon, Sep 30, 2019 at 3:13 PM Josh Hunt <joh...@akamai.com> wrote:
> >
> > Prior to this change an application sending <= 1MSS worth of data and
> > enabling UDP GSO would fail if the system had SW GSO enabled, but the
> > same send would succeed if HW GSO offload is enabled. In addition to this
> > inconsistency the error in the SW GSO case does not get back to the
> > application if sending out of a real device so the user is unaware of this
> > failure.
> >
> > With this change we only perform GSO if the # of segments is > 1 even
> > if the application has enabled segmentation. I've also updated the
> > relevant udpgso selftests.
> >
> > Fixes: bec1f6f69736 ("udp: generate gso with UDP_SEGMENT")
> > Signed-off-by: Josh Hunt <joh...@akamai.com>
> > ---
> > net/ipv4/udp.c | 5 +++--
> > net/ipv6/udp.c | 5 +++--
> > tools/testing/selftests/net/udpgso.c | 16 ++++------------
> > 3 files changed, 10 insertions(+), 16 deletions(-)
> >
> > diff --git a/net/ipv4/udp.c b/net/ipv4/udp.c
> > index be98d0b8f014..ac0baf947560 100644
> > --- a/net/ipv4/udp.c
> > +++ b/net/ipv4/udp.c
> > @@ -821,6 +821,7 @@ static int udp_send_skb(struct sk_buff *skb, struct
> > flowi4 *fl4,
> > int is_udplite = IS_UDPLITE(sk);
> > int offset = skb_transport_offset(skb);
> > int len = skb->len - offset;
> > + int datalen = len - sizeof(*uh);
> > __wsum csum = 0;
> >
> > /*
> > @@ -832,7 +833,7 @@ static int udp_send_skb(struct sk_buff *skb, struct
> > flowi4 *fl4,
> > uh->len = htons(len);
> > uh->check = 0;
> >
> > - if (cork->gso_size) {
> > + if (cork->gso_size && datalen > cork->gso_size) {
> > const int hlen = skb_network_header_len(skb) +
> > sizeof(struct udphdr);
> >
>
> So what about the datalen == cork->gso_size case? That would only
> generate one segment wouldn't it?
Segmentation drops packets in this boundary case (not sure why).
> Shouldn't the test really be "datalen < cork->gso_size"? That should
> be the only check you need since if gso_size is 0 this statement would
> always fail anyway.
>
> Thanks.
>
> - Alex
The original choice was made to match GSO behavior of other protocols.
The drop occurs in protocol-independent skb_segment.
But I had not anticipated HW GSO to behave differently. With that,
aligning the two makes sense. Especially as UDP GSO is exposed to
userspace. Having to explicitly code a branch whether or not to pass
UDP_SEGMENT on each send based on size is confusing.
gso_size is supplied by the user. That value need not be smaller than
or equal to MTU minus headers. Some of the tests inside the branch,
especially
if (hlen + cork->gso_size > cork->fragsize) {
kfree_skb(skb);
return -EINVAL;
}
still need to be checked.
