On Thu, 2016-12-22 at 16:26 -0500, Josef Bacik wrote:
> If we have non reuseport sockets on a tb we will set tb->fastreuseport to 0
> and
> never set it again. Which means that in the future if we end up adding a
> bunch
> of reuseport sk's to that tb we'll have to do the expensive scan every time.
> Instead add a sock_common to the tb so we know what reuseport sk succeeded
> last.
> Once one sk has made it onto the list we know that there are no potential bind
> conflicts on the owners list that match that sk's rcv_addr. So copy the sk's
> common into our tb->fastsock and set tb->fastruseport to FASTREUSESOCK_STRICT
> so
> we know we have to do an extra check for subsequent reuseport sockets and skip
> the expensive bind conflict check.
>
> Signed-off-by: Josef Bacik <jba...@fb.com>
> ---
> include/net/inet_hashtables.h | 4 ++++
> net/ipv4/inet_connection_sock.c | 53
> +++++++++++++++++++++++++++++++++++++----
> 2 files changed, 53 insertions(+), 4 deletions(-)
>
> diff --git a/include/net/inet_hashtables.h b/include/net/inet_hashtables.h
> index 756ed16..4ccc18f 100644
> --- a/include/net/inet_hashtables.h
> +++ b/include/net/inet_hashtables.h
> @@ -74,12 +74,16 @@ struct inet_ehash_bucket {
> * users logged onto your box, isn't it nice to know that new data
> * ports are created in O(1) time? I thought so. ;-) -DaveM
> */
> +#define FASTREUSEPORT_ANY 1
> +#define FASTREUSEPORT_STRICT 2
> +
> struct inet_bind_bucket {
> possible_net_t ib_net;
> unsigned short port;
> signed char fastreuse;
> signed char fastreuseport;
> kuid_t fastuid;
> + struct sock_common fastsock;
> int num_owners;
> struct hlist_node node;
> struct hlist_head owners;
Please place this fat field at the end of inet_bind_bucket
Many sockets do not use SO_REUSEPORT and should not use this field,
while tb->owners need to be touched.
