tc/htb: Hierarchical QoS hardware offload

tc/htb: Hierarchical QoS hardware offload

This commit adds support for configuring HTB in offload mode. HTB
offload eliminates the single qdisc lock in the datapath and offloads
the algorithm to the NIC. The new 'offload' parameter is added to
enable this mode:

    # tc qdisc replace dev eth0 root handle 1: htb offload

Classes are created as usual, but filters should be moved to clsact for
lock-free classification (filters attached to HTB itself are not
supported in the offload mode):

    # tc filter add dev eth0 egress protocol ip flower dst_port 80
    action skbedit priority 1:10

tc qdisc show and tc class show will indicate whether the offload is
enabled. Example output:

$ tc qdisc show dev eth1
qdisc htb 1: root offloaded r2q 10 default 0 direct_packets_stat 0 direct_qlen 1000 offload
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
qdisc pfifo 0: parent 1: limit 1000p
$ tc class show dev eth1
class htb 1:101 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:1 root rate 100Gbit ceil 100Gbit burst 0b cburst 0b  offload
class htb 1:103 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:102 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:105 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:104 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:107 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:106 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
class htb 1:108 parent 1:1 prio 0 rate 4Gbit ceil 4Gbit burst 1000b cburst 1000b  offload
$ tc -j qdisc show dev eth1
[{"kind":"htb","handle":"1:","root":true,"offloaded":true,"options":{"r2q":10,"default":"0","direct_packets_stat":0,"direct_qlen":1000,"offload":null}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}},{"kind":"pfifo","handle":"0:","parent":"1:","options":{"limit":1000}}]

Signed-off-by: Maxim Mikityanskiy <maximmi@mellanox.com>
Reviewed-by: Tariq Toukan <tariqt@nvidia.com>
Signed-off-by: David Ahern <dsahern@kernel.org>