mirror of
https://github.com/openvswitch/ovs
synced 2025-10-13 14:07:02 +00:00
Until now, knowledge about OpenFlow has been somewhat scattered around the tree. Some of it is in ofp-actions, some of it is in ofp-util, some in separate files for individual actions, and most of the wire format declarations are in include/openflow. This commit centralizes all of that in ofp-actions. Encoding and decoding OpenFlow actions was previously broken up by OpenFlow version. This was OK with only OpenFlow 1.0 and 1.1, but each additional version added a new wrapper around the existing ones, which started to become hard to understand. This commit merges all of the processing for the different versions, to the extent that they are similar, making the version differences clearer. Previously, ofp-actions contained OpenFlow encoding and decoding, plus ofpact formatting, but OpenFlow parsing was separated into ofp-parse, which seems an odd division. This commit moves the parsing code into ofp-actions with the rest of the code. Before this commit, the four main bits of code associated with a particular ofpact--OpenFlow encoding and decoding, ofpact formatting and parsing--were all found far away from each other. This often made it hard to see what was going on for a particular ofpact, since you had to search around to many different pieces of code. This commit reorganizes so that all of the code for a given ofpact is in a single place. As a code refactoring, this commit has little visible behavioral change. The update to ofproto-dpif.at illustrates one minor bug fix as a side effect: a flow that was added with the action "dec_ttl" (a standard OpenFlow action) was previously formatted as "dec_ttl(0)" (using a Nicira extension to specifically direct packets bounced to the controller because of too-low TTL), but after this commit it is correctly formatted as "dec_ttl". The other visible effect is to drop support for the Nicira extension dec_ttl action in OpenFlow 1.1 and later in favor of the equivalent standard action. It seems unlikely that anyone was really using the Nicira extension in OF1.1 or later. Signed-off-by: Ben Pfaff <blp@nicira.com> Acked-by: Jarno Rajahalme <jrajahalme@nicira.com>
7908 lines
252 KiB
C
7908 lines
252 KiB
C
/*
|
||
* Copyright (c) 2008, 2009, 2010, 2011, 2012, 2013, 2014 Nicira, Inc.
|
||
*
|
||
* Licensed under the Apache License, Version 2.0 (the "License");
|
||
* you may not use this file except in compliance with the License.
|
||
* You may obtain a copy of the License at:
|
||
*
|
||
* http://www.apache.org/licenses/LICENSE-2.0
|
||
*
|
||
* Unless required by applicable law or agreed to in writing, software
|
||
* distributed under the License is distributed on an "AS IS" BASIS,
|
||
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
|
||
* See the License for the specific language governing permissions and
|
||
* limitations under the License.
|
||
*/
|
||
|
||
#include <config.h>
|
||
#include "ofp-print.h"
|
||
#include <ctype.h>
|
||
#include <errno.h>
|
||
#include <inttypes.h>
|
||
#include <sys/types.h>
|
||
#include <netinet/in.h>
|
||
#include <netinet/icmp6.h>
|
||
#include <stdlib.h>
|
||
#include "bundle.h"
|
||
#include "byte-order.h"
|
||
#include "classifier.h"
|
||
#include "dynamic-string.h"
|
||
#include "learn.h"
|
||
#include "meta-flow.h"
|
||
#include "multipath.h"
|
||
#include "netdev.h"
|
||
#include "nx-match.h"
|
||
#include "ofp-actions.h"
|
||
#include "ofp-errors.h"
|
||
#include "ofp-msgs.h"
|
||
#include "ofp-util.h"
|
||
#include "ofpbuf.h"
|
||
#include "packets.h"
|
||
#include "random.h"
|
||
#include "unaligned.h"
|
||
#include "type-props.h"
|
||
#include "vlog.h"
|
||
#include "bitmap.h"
|
||
|
||
VLOG_DEFINE_THIS_MODULE(ofp_util);
|
||
|
||
/* Rate limit for OpenFlow message parse errors. These always indicate a bug
|
||
* in the peer and so there's not much point in showing a lot of them. */
|
||
static struct vlog_rate_limit bad_ofmsg_rl = VLOG_RATE_LIMIT_INIT(1, 5);
|
||
|
||
static enum ofputil_table_miss ofputil_table_miss_from_config(
|
||
ovs_be32 config_, enum ofp_version);
|
||
|
||
struct ofp_prop_header {
|
||
ovs_be16 type;
|
||
ovs_be16 len;
|
||
};
|
||
|
||
/* Pulls a property, beginning with struct ofp_prop_header, from the beginning
|
||
* of 'msg'. Stores the type of the property in '*typep' and, if 'property' is
|
||
* nonnull, the entire property, including the header, in '*property'. Returns
|
||
* 0 if successful, otherwise an error code.
|
||
*
|
||
* This function pulls the property's stated size padded out to a multiple of
|
||
* 'alignment' bytes. The common case in OpenFlow is an 'alignment' of 8, so
|
||
* you can use ofputil_pull_property() for that case. */
|
||
static enum ofperr
|
||
ofputil_pull_property__(struct ofpbuf *msg, struct ofpbuf *property,
|
||
unsigned int alignment, uint16_t *typep)
|
||
{
|
||
struct ofp_prop_header *oph;
|
||
unsigned int padded_len;
|
||
unsigned int len;
|
||
|
||
if (ofpbuf_size(msg) < sizeof *oph) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
|
||
oph = ofpbuf_data(msg);
|
||
len = ntohs(oph->len);
|
||
padded_len = ROUND_UP(len, alignment);
|
||
if (len < sizeof *oph || padded_len > ofpbuf_size(msg)) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
|
||
*typep = ntohs(oph->type);
|
||
if (property) {
|
||
ofpbuf_use_const(property, ofpbuf_data(msg), len);
|
||
}
|
||
ofpbuf_pull(msg, padded_len);
|
||
return 0;
|
||
}
|
||
|
||
/* Pulls a property, beginning with struct ofp_prop_header, from the beginning
|
||
* of 'msg'. Stores the type of the property in '*typep' and, if 'property' is
|
||
* nonnull, the entire property, including the header, in '*property'. Returns
|
||
* 0 if successful, otherwise an error code.
|
||
*
|
||
* This function pulls the property's stated size padded out to a multiple of
|
||
* 8 bytes, which is the common case for OpenFlow properties. */
|
||
static enum ofperr
|
||
ofputil_pull_property(struct ofpbuf *msg, struct ofpbuf *property,
|
||
uint16_t *typep)
|
||
{
|
||
return ofputil_pull_property__(msg, property, 8, typep);
|
||
}
|
||
|
||
static void PRINTF_FORMAT(2, 3)
|
||
log_property(bool loose, const char *message, ...)
|
||
{
|
||
enum vlog_level level = loose ? VLL_DBG : VLL_WARN;
|
||
if (!vlog_should_drop(THIS_MODULE, level, &bad_ofmsg_rl)) {
|
||
va_list args;
|
||
|
||
va_start(args, message);
|
||
vlog_valist(THIS_MODULE, level, message, args);
|
||
va_end(args);
|
||
}
|
||
}
|
||
|
||
static size_t
|
||
start_property(struct ofpbuf *msg, uint16_t type)
|
||
{
|
||
size_t start_ofs = ofpbuf_size(msg);
|
||
struct ofp_prop_header *oph;
|
||
|
||
oph = ofpbuf_put_uninit(msg, sizeof *oph);
|
||
oph->type = htons(type);
|
||
oph->len = htons(4); /* May be updated later by end_property(). */
|
||
return start_ofs;
|
||
}
|
||
|
||
static void
|
||
end_property(struct ofpbuf *msg, size_t start_ofs)
|
||
{
|
||
struct ofp_prop_header *oph;
|
||
|
||
oph = ofpbuf_at_assert(msg, start_ofs, sizeof *oph);
|
||
oph->len = htons(ofpbuf_size(msg) - start_ofs);
|
||
ofpbuf_padto(msg, ROUND_UP(ofpbuf_size(msg), 8));
|
||
}
|
||
|
||
static void
|
||
put_bitmap_properties(struct ofpbuf *msg, uint64_t bitmap)
|
||
{
|
||
for (; bitmap; bitmap = zero_rightmost_1bit(bitmap)) {
|
||
start_property(msg, rightmost_1bit_idx(bitmap));
|
||
}
|
||
}
|
||
|
||
/* Given the wildcard bit count in the least-significant 6 of 'wcbits', returns
|
||
* an IP netmask with a 1 in each bit that must match and a 0 in each bit that
|
||
* is wildcarded.
|
||
*
|
||
* The bits in 'wcbits' are in the format used in enum ofp_flow_wildcards: 0
|
||
* is exact match, 1 ignores the LSB, 2 ignores the 2 least-significant bits,
|
||
* ..., 32 and higher wildcard the entire field. This is the *opposite* of the
|
||
* usual convention where e.g. /24 indicates that 8 bits (not 24 bits) are
|
||
* wildcarded. */
|
||
ovs_be32
|
||
ofputil_wcbits_to_netmask(int wcbits)
|
||
{
|
||
wcbits &= 0x3f;
|
||
return wcbits < 32 ? htonl(~((1u << wcbits) - 1)) : 0;
|
||
}
|
||
|
||
/* Given the IP netmask 'netmask', returns the number of bits of the IP address
|
||
* that it wildcards, that is, the number of 0-bits in 'netmask', a number
|
||
* between 0 and 32 inclusive.
|
||
*
|
||
* If 'netmask' is not a CIDR netmask (see ip_is_cidr()), the return value will
|
||
* still be in the valid range but isn't otherwise meaningful. */
|
||
int
|
||
ofputil_netmask_to_wcbits(ovs_be32 netmask)
|
||
{
|
||
return 32 - ip_count_cidr_bits(netmask);
|
||
}
|
||
|
||
/* Converts the OpenFlow 1.0 wildcards in 'ofpfw' (OFPFW10_*) into a
|
||
* flow_wildcards in 'wc' for use in struct match. It is the caller's
|
||
* responsibility to handle the special case where the flow match's dl_vlan is
|
||
* set to OFP_VLAN_NONE. */
|
||
void
|
||
ofputil_wildcard_from_ofpfw10(uint32_t ofpfw, struct flow_wildcards *wc)
|
||
{
|
||
BUILD_ASSERT_DECL(FLOW_WC_SEQ == 27);
|
||
|
||
/* Initialize most of wc. */
|
||
flow_wildcards_init_catchall(wc);
|
||
|
||
if (!(ofpfw & OFPFW10_IN_PORT)) {
|
||
wc->masks.in_port.ofp_port = u16_to_ofp(UINT16_MAX);
|
||
}
|
||
|
||
if (!(ofpfw & OFPFW10_NW_TOS)) {
|
||
wc->masks.nw_tos |= IP_DSCP_MASK;
|
||
}
|
||
|
||
if (!(ofpfw & OFPFW10_NW_PROTO)) {
|
||
wc->masks.nw_proto = UINT8_MAX;
|
||
}
|
||
wc->masks.nw_src = ofputil_wcbits_to_netmask(ofpfw
|
||
>> OFPFW10_NW_SRC_SHIFT);
|
||
wc->masks.nw_dst = ofputil_wcbits_to_netmask(ofpfw
|
||
>> OFPFW10_NW_DST_SHIFT);
|
||
|
||
if (!(ofpfw & OFPFW10_TP_SRC)) {
|
||
wc->masks.tp_src = OVS_BE16_MAX;
|
||
}
|
||
if (!(ofpfw & OFPFW10_TP_DST)) {
|
||
wc->masks.tp_dst = OVS_BE16_MAX;
|
||
}
|
||
|
||
if (!(ofpfw & OFPFW10_DL_SRC)) {
|
||
memset(wc->masks.dl_src, 0xff, ETH_ADDR_LEN);
|
||
}
|
||
if (!(ofpfw & OFPFW10_DL_DST)) {
|
||
memset(wc->masks.dl_dst, 0xff, ETH_ADDR_LEN);
|
||
}
|
||
if (!(ofpfw & OFPFW10_DL_TYPE)) {
|
||
wc->masks.dl_type = OVS_BE16_MAX;
|
||
}
|
||
|
||
/* VLAN TCI mask. */
|
||
if (!(ofpfw & OFPFW10_DL_VLAN_PCP)) {
|
||
wc->masks.vlan_tci |= htons(VLAN_PCP_MASK | VLAN_CFI);
|
||
}
|
||
if (!(ofpfw & OFPFW10_DL_VLAN)) {
|
||
wc->masks.vlan_tci |= htons(VLAN_VID_MASK | VLAN_CFI);
|
||
}
|
||
}
|
||
|
||
/* Converts the ofp10_match in 'ofmatch' into a struct match in 'match'. */
|
||
void
|
||
ofputil_match_from_ofp10_match(const struct ofp10_match *ofmatch,
|
||
struct match *match)
|
||
{
|
||
uint32_t ofpfw = ntohl(ofmatch->wildcards) & OFPFW10_ALL;
|
||
|
||
/* Initialize match->wc. */
|
||
memset(&match->flow, 0, sizeof match->flow);
|
||
ofputil_wildcard_from_ofpfw10(ofpfw, &match->wc);
|
||
|
||
/* Initialize most of match->flow. */
|
||
match->flow.nw_src = ofmatch->nw_src;
|
||
match->flow.nw_dst = ofmatch->nw_dst;
|
||
match->flow.in_port.ofp_port = u16_to_ofp(ntohs(ofmatch->in_port));
|
||
match->flow.dl_type = ofputil_dl_type_from_openflow(ofmatch->dl_type);
|
||
match->flow.tp_src = ofmatch->tp_src;
|
||
match->flow.tp_dst = ofmatch->tp_dst;
|
||
memcpy(match->flow.dl_src, ofmatch->dl_src, ETH_ADDR_LEN);
|
||
memcpy(match->flow.dl_dst, ofmatch->dl_dst, ETH_ADDR_LEN);
|
||
match->flow.nw_tos = ofmatch->nw_tos & IP_DSCP_MASK;
|
||
match->flow.nw_proto = ofmatch->nw_proto;
|
||
|
||
/* Translate VLANs. */
|
||
if (!(ofpfw & OFPFW10_DL_VLAN) &&
|
||
ofmatch->dl_vlan == htons(OFP10_VLAN_NONE)) {
|
||
/* Match only packets without 802.1Q header.
|
||
*
|
||
* When OFPFW10_DL_VLAN_PCP is wildcarded, this is obviously correct.
|
||
*
|
||
* If OFPFW10_DL_VLAN_PCP is matched, the flow match is contradictory,
|
||
* because we can't have a specific PCP without an 802.1Q header.
|
||
* However, older versions of OVS treated this as matching packets
|
||
* withut an 802.1Q header, so we do here too. */
|
||
match->flow.vlan_tci = htons(0);
|
||
match->wc.masks.vlan_tci = htons(0xffff);
|
||
} else {
|
||
ovs_be16 vid, pcp, tci;
|
||
uint16_t hpcp;
|
||
|
||
vid = ofmatch->dl_vlan & htons(VLAN_VID_MASK);
|
||
hpcp = (ofmatch->dl_vlan_pcp << VLAN_PCP_SHIFT) & VLAN_PCP_MASK;
|
||
pcp = htons(hpcp);
|
||
tci = vid | pcp | htons(VLAN_CFI);
|
||
match->flow.vlan_tci = tci & match->wc.masks.vlan_tci;
|
||
}
|
||
|
||
/* Clean up. */
|
||
match_zero_wildcarded_fields(match);
|
||
}
|
||
|
||
/* Convert 'match' into the OpenFlow 1.0 match structure 'ofmatch'. */
|
||
void
|
||
ofputil_match_to_ofp10_match(const struct match *match,
|
||
struct ofp10_match *ofmatch)
|
||
{
|
||
const struct flow_wildcards *wc = &match->wc;
|
||
uint32_t ofpfw;
|
||
|
||
/* Figure out most OpenFlow wildcards. */
|
||
ofpfw = 0;
|
||
if (!wc->masks.in_port.ofp_port) {
|
||
ofpfw |= OFPFW10_IN_PORT;
|
||
}
|
||
if (!wc->masks.dl_type) {
|
||
ofpfw |= OFPFW10_DL_TYPE;
|
||
}
|
||
if (!wc->masks.nw_proto) {
|
||
ofpfw |= OFPFW10_NW_PROTO;
|
||
}
|
||
ofpfw |= (ofputil_netmask_to_wcbits(wc->masks.nw_src)
|
||
<< OFPFW10_NW_SRC_SHIFT);
|
||
ofpfw |= (ofputil_netmask_to_wcbits(wc->masks.nw_dst)
|
||
<< OFPFW10_NW_DST_SHIFT);
|
||
if (!(wc->masks.nw_tos & IP_DSCP_MASK)) {
|
||
ofpfw |= OFPFW10_NW_TOS;
|
||
}
|
||
if (!wc->masks.tp_src) {
|
||
ofpfw |= OFPFW10_TP_SRC;
|
||
}
|
||
if (!wc->masks.tp_dst) {
|
||
ofpfw |= OFPFW10_TP_DST;
|
||
}
|
||
if (eth_addr_is_zero(wc->masks.dl_src)) {
|
||
ofpfw |= OFPFW10_DL_SRC;
|
||
}
|
||
if (eth_addr_is_zero(wc->masks.dl_dst)) {
|
||
ofpfw |= OFPFW10_DL_DST;
|
||
}
|
||
|
||
/* Translate VLANs. */
|
||
ofmatch->dl_vlan = htons(0);
|
||
ofmatch->dl_vlan_pcp = 0;
|
||
if (match->wc.masks.vlan_tci == htons(0)) {
|
||
ofpfw |= OFPFW10_DL_VLAN | OFPFW10_DL_VLAN_PCP;
|
||
} else if (match->wc.masks.vlan_tci & htons(VLAN_CFI)
|
||
&& !(match->flow.vlan_tci & htons(VLAN_CFI))) {
|
||
ofmatch->dl_vlan = htons(OFP10_VLAN_NONE);
|
||
ofpfw |= OFPFW10_DL_VLAN_PCP;
|
||
} else {
|
||
if (!(match->wc.masks.vlan_tci & htons(VLAN_VID_MASK))) {
|
||
ofpfw |= OFPFW10_DL_VLAN;
|
||
} else {
|
||
ofmatch->dl_vlan = htons(vlan_tci_to_vid(match->flow.vlan_tci));
|
||
}
|
||
|
||
if (!(match->wc.masks.vlan_tci & htons(VLAN_PCP_MASK))) {
|
||
ofpfw |= OFPFW10_DL_VLAN_PCP;
|
||
} else {
|
||
ofmatch->dl_vlan_pcp = vlan_tci_to_pcp(match->flow.vlan_tci);
|
||
}
|
||
}
|
||
|
||
/* Compose most of the match structure. */
|
||
ofmatch->wildcards = htonl(ofpfw);
|
||
ofmatch->in_port = htons(ofp_to_u16(match->flow.in_port.ofp_port));
|
||
memcpy(ofmatch->dl_src, match->flow.dl_src, ETH_ADDR_LEN);
|
||
memcpy(ofmatch->dl_dst, match->flow.dl_dst, ETH_ADDR_LEN);
|
||
ofmatch->dl_type = ofputil_dl_type_to_openflow(match->flow.dl_type);
|
||
ofmatch->nw_src = match->flow.nw_src;
|
||
ofmatch->nw_dst = match->flow.nw_dst;
|
||
ofmatch->nw_tos = match->flow.nw_tos & IP_DSCP_MASK;
|
||
ofmatch->nw_proto = match->flow.nw_proto;
|
||
ofmatch->tp_src = match->flow.tp_src;
|
||
ofmatch->tp_dst = match->flow.tp_dst;
|
||
memset(ofmatch->pad1, '\0', sizeof ofmatch->pad1);
|
||
memset(ofmatch->pad2, '\0', sizeof ofmatch->pad2);
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_pull_ofp11_match(struct ofpbuf *buf, struct match *match,
|
||
uint16_t *padded_match_len)
|
||
{
|
||
struct ofp11_match_header *omh = ofpbuf_data(buf);
|
||
uint16_t match_len;
|
||
|
||
if (ofpbuf_size(buf) < sizeof *omh) {
|
||
return OFPERR_OFPBMC_BAD_LEN;
|
||
}
|
||
|
||
match_len = ntohs(omh->length);
|
||
|
||
switch (ntohs(omh->type)) {
|
||
case OFPMT_STANDARD: {
|
||
struct ofp11_match *om;
|
||
|
||
if (match_len != sizeof *om || ofpbuf_size(buf) < sizeof *om) {
|
||
return OFPERR_OFPBMC_BAD_LEN;
|
||
}
|
||
om = ofpbuf_pull(buf, sizeof *om);
|
||
if (padded_match_len) {
|
||
*padded_match_len = match_len;
|
||
}
|
||
return ofputil_match_from_ofp11_match(om, match);
|
||
}
|
||
|
||
case OFPMT_OXM:
|
||
if (padded_match_len) {
|
||
*padded_match_len = ROUND_UP(match_len, 8);
|
||
}
|
||
return oxm_pull_match(buf, match);
|
||
|
||
default:
|
||
return OFPERR_OFPBMC_BAD_TYPE;
|
||
}
|
||
}
|
||
|
||
/* Converts the ofp11_match in 'ofmatch' into a struct match in 'match'.
|
||
* Returns 0 if successful, otherwise an OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_match_from_ofp11_match(const struct ofp11_match *ofmatch,
|
||
struct match *match)
|
||
{
|
||
uint16_t wc = ntohl(ofmatch->wildcards);
|
||
uint8_t dl_src_mask[ETH_ADDR_LEN];
|
||
uint8_t dl_dst_mask[ETH_ADDR_LEN];
|
||
bool ipv4, arp, rarp;
|
||
int i;
|
||
|
||
match_init_catchall(match);
|
||
|
||
if (!(wc & OFPFW11_IN_PORT)) {
|
||
ofp_port_t ofp_port;
|
||
enum ofperr error;
|
||
|
||
error = ofputil_port_from_ofp11(ofmatch->in_port, &ofp_port);
|
||
if (error) {
|
||
return OFPERR_OFPBMC_BAD_VALUE;
|
||
}
|
||
match_set_in_port(match, ofp_port);
|
||
}
|
||
|
||
for (i = 0; i < ETH_ADDR_LEN; i++) {
|
||
dl_src_mask[i] = ~ofmatch->dl_src_mask[i];
|
||
}
|
||
match_set_dl_src_masked(match, ofmatch->dl_src, dl_src_mask);
|
||
|
||
for (i = 0; i < ETH_ADDR_LEN; i++) {
|
||
dl_dst_mask[i] = ~ofmatch->dl_dst_mask[i];
|
||
}
|
||
match_set_dl_dst_masked(match, ofmatch->dl_dst, dl_dst_mask);
|
||
|
||
if (!(wc & OFPFW11_DL_VLAN)) {
|
||
if (ofmatch->dl_vlan == htons(OFPVID11_NONE)) {
|
||
/* Match only packets without a VLAN tag. */
|
||
match->flow.vlan_tci = htons(0);
|
||
match->wc.masks.vlan_tci = OVS_BE16_MAX;
|
||
} else {
|
||
if (ofmatch->dl_vlan == htons(OFPVID11_ANY)) {
|
||
/* Match any packet with a VLAN tag regardless of VID. */
|
||
match->flow.vlan_tci = htons(VLAN_CFI);
|
||
match->wc.masks.vlan_tci = htons(VLAN_CFI);
|
||
} else if (ntohs(ofmatch->dl_vlan) < 4096) {
|
||
/* Match only packets with the specified VLAN VID. */
|
||
match->flow.vlan_tci = htons(VLAN_CFI) | ofmatch->dl_vlan;
|
||
match->wc.masks.vlan_tci = htons(VLAN_CFI | VLAN_VID_MASK);
|
||
} else {
|
||
/* Invalid VID. */
|
||
return OFPERR_OFPBMC_BAD_VALUE;
|
||
}
|
||
|
||
if (!(wc & OFPFW11_DL_VLAN_PCP)) {
|
||
if (ofmatch->dl_vlan_pcp <= 7) {
|
||
match->flow.vlan_tci |= htons(ofmatch->dl_vlan_pcp
|
||
<< VLAN_PCP_SHIFT);
|
||
match->wc.masks.vlan_tci |= htons(VLAN_PCP_MASK);
|
||
} else {
|
||
/* Invalid PCP. */
|
||
return OFPERR_OFPBMC_BAD_VALUE;
|
||
}
|
||
}
|
||
}
|
||
}
|
||
|
||
if (!(wc & OFPFW11_DL_TYPE)) {
|
||
match_set_dl_type(match,
|
||
ofputil_dl_type_from_openflow(ofmatch->dl_type));
|
||
}
|
||
|
||
ipv4 = match->flow.dl_type == htons(ETH_TYPE_IP);
|
||
arp = match->flow.dl_type == htons(ETH_TYPE_ARP);
|
||
rarp = match->flow.dl_type == htons(ETH_TYPE_RARP);
|
||
|
||
if (ipv4 && !(wc & OFPFW11_NW_TOS)) {
|
||
if (ofmatch->nw_tos & ~IP_DSCP_MASK) {
|
||
/* Invalid TOS. */
|
||
return OFPERR_OFPBMC_BAD_VALUE;
|
||
}
|
||
|
||
match_set_nw_dscp(match, ofmatch->nw_tos);
|
||
}
|
||
|
||
if (ipv4 || arp || rarp) {
|
||
if (!(wc & OFPFW11_NW_PROTO)) {
|
||
match_set_nw_proto(match, ofmatch->nw_proto);
|
||
}
|
||
match_set_nw_src_masked(match, ofmatch->nw_src, ~ofmatch->nw_src_mask);
|
||
match_set_nw_dst_masked(match, ofmatch->nw_dst, ~ofmatch->nw_dst_mask);
|
||
}
|
||
|
||
#define OFPFW11_TP_ALL (OFPFW11_TP_SRC | OFPFW11_TP_DST)
|
||
if (ipv4 && (wc & OFPFW11_TP_ALL) != OFPFW11_TP_ALL) {
|
||
switch (match->flow.nw_proto) {
|
||
case IPPROTO_ICMP:
|
||
/* "A.2.3 Flow Match Structures" in OF1.1 says:
|
||
*
|
||
* The tp_src and tp_dst fields will be ignored unless the
|
||
* network protocol specified is as TCP, UDP or SCTP.
|
||
*
|
||
* but I'm pretty sure we should support ICMP too, otherwise
|
||
* that's a regression from OF1.0. */
|
||
if (!(wc & OFPFW11_TP_SRC)) {
|
||
uint16_t icmp_type = ntohs(ofmatch->tp_src);
|
||
if (icmp_type < 0x100) {
|
||
match_set_icmp_type(match, icmp_type);
|
||
} else {
|
||
return OFPERR_OFPBMC_BAD_FIELD;
|
||
}
|
||
}
|
||
if (!(wc & OFPFW11_TP_DST)) {
|
||
uint16_t icmp_code = ntohs(ofmatch->tp_dst);
|
||
if (icmp_code < 0x100) {
|
||
match_set_icmp_code(match, icmp_code);
|
||
} else {
|
||
return OFPERR_OFPBMC_BAD_FIELD;
|
||
}
|
||
}
|
||
break;
|
||
|
||
case IPPROTO_TCP:
|
||
case IPPROTO_UDP:
|
||
case IPPROTO_SCTP:
|
||
if (!(wc & (OFPFW11_TP_SRC))) {
|
||
match_set_tp_src(match, ofmatch->tp_src);
|
||
}
|
||
if (!(wc & (OFPFW11_TP_DST))) {
|
||
match_set_tp_dst(match, ofmatch->tp_dst);
|
||
}
|
||
break;
|
||
|
||
default:
|
||
/* OF1.1 says explicitly to ignore this. */
|
||
break;
|
||
}
|
||
}
|
||
|
||
if (eth_type_mpls(match->flow.dl_type)) {
|
||
if (!(wc & OFPFW11_MPLS_LABEL)) {
|
||
match_set_mpls_label(match, 0, ofmatch->mpls_label);
|
||
}
|
||
if (!(wc & OFPFW11_MPLS_TC)) {
|
||
match_set_mpls_tc(match, 0, ofmatch->mpls_tc);
|
||
}
|
||
}
|
||
|
||
match_set_metadata_masked(match, ofmatch->metadata,
|
||
~ofmatch->metadata_mask);
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Convert 'match' into the OpenFlow 1.1 match structure 'ofmatch'. */
|
||
void
|
||
ofputil_match_to_ofp11_match(const struct match *match,
|
||
struct ofp11_match *ofmatch)
|
||
{
|
||
uint32_t wc = 0;
|
||
int i;
|
||
|
||
memset(ofmatch, 0, sizeof *ofmatch);
|
||
ofmatch->omh.type = htons(OFPMT_STANDARD);
|
||
ofmatch->omh.length = htons(OFPMT11_STANDARD_LENGTH);
|
||
|
||
if (!match->wc.masks.in_port.ofp_port) {
|
||
wc |= OFPFW11_IN_PORT;
|
||
} else {
|
||
ofmatch->in_port = ofputil_port_to_ofp11(match->flow.in_port.ofp_port);
|
||
}
|
||
|
||
memcpy(ofmatch->dl_src, match->flow.dl_src, ETH_ADDR_LEN);
|
||
for (i = 0; i < ETH_ADDR_LEN; i++) {
|
||
ofmatch->dl_src_mask[i] = ~match->wc.masks.dl_src[i];
|
||
}
|
||
|
||
memcpy(ofmatch->dl_dst, match->flow.dl_dst, ETH_ADDR_LEN);
|
||
for (i = 0; i < ETH_ADDR_LEN; i++) {
|
||
ofmatch->dl_dst_mask[i] = ~match->wc.masks.dl_dst[i];
|
||
}
|
||
|
||
if (match->wc.masks.vlan_tci == htons(0)) {
|
||
wc |= OFPFW11_DL_VLAN | OFPFW11_DL_VLAN_PCP;
|
||
} else if (match->wc.masks.vlan_tci & htons(VLAN_CFI)
|
||
&& !(match->flow.vlan_tci & htons(VLAN_CFI))) {
|
||
ofmatch->dl_vlan = htons(OFPVID11_NONE);
|
||
wc |= OFPFW11_DL_VLAN_PCP;
|
||
} else {
|
||
if (!(match->wc.masks.vlan_tci & htons(VLAN_VID_MASK))) {
|
||
ofmatch->dl_vlan = htons(OFPVID11_ANY);
|
||
} else {
|
||
ofmatch->dl_vlan = htons(vlan_tci_to_vid(match->flow.vlan_tci));
|
||
}
|
||
|
||
if (!(match->wc.masks.vlan_tci & htons(VLAN_PCP_MASK))) {
|
||
wc |= OFPFW11_DL_VLAN_PCP;
|
||
} else {
|
||
ofmatch->dl_vlan_pcp = vlan_tci_to_pcp(match->flow.vlan_tci);
|
||
}
|
||
}
|
||
|
||
if (!match->wc.masks.dl_type) {
|
||
wc |= OFPFW11_DL_TYPE;
|
||
} else {
|
||
ofmatch->dl_type = ofputil_dl_type_to_openflow(match->flow.dl_type);
|
||
}
|
||
|
||
if (!(match->wc.masks.nw_tos & IP_DSCP_MASK)) {
|
||
wc |= OFPFW11_NW_TOS;
|
||
} else {
|
||
ofmatch->nw_tos = match->flow.nw_tos & IP_DSCP_MASK;
|
||
}
|
||
|
||
if (!match->wc.masks.nw_proto) {
|
||
wc |= OFPFW11_NW_PROTO;
|
||
} else {
|
||
ofmatch->nw_proto = match->flow.nw_proto;
|
||
}
|
||
|
||
ofmatch->nw_src = match->flow.nw_src;
|
||
ofmatch->nw_src_mask = ~match->wc.masks.nw_src;
|
||
ofmatch->nw_dst = match->flow.nw_dst;
|
||
ofmatch->nw_dst_mask = ~match->wc.masks.nw_dst;
|
||
|
||
if (!match->wc.masks.tp_src) {
|
||
wc |= OFPFW11_TP_SRC;
|
||
} else {
|
||
ofmatch->tp_src = match->flow.tp_src;
|
||
}
|
||
|
||
if (!match->wc.masks.tp_dst) {
|
||
wc |= OFPFW11_TP_DST;
|
||
} else {
|
||
ofmatch->tp_dst = match->flow.tp_dst;
|
||
}
|
||
|
||
if (!(match->wc.masks.mpls_lse[0] & htonl(MPLS_LABEL_MASK))) {
|
||
wc |= OFPFW11_MPLS_LABEL;
|
||
} else {
|
||
ofmatch->mpls_label = htonl(mpls_lse_to_label(
|
||
match->flow.mpls_lse[0]));
|
||
}
|
||
|
||
if (!(match->wc.masks.mpls_lse[0] & htonl(MPLS_TC_MASK))) {
|
||
wc |= OFPFW11_MPLS_TC;
|
||
} else {
|
||
ofmatch->mpls_tc = mpls_lse_to_tc(match->flow.mpls_lse[0]);
|
||
}
|
||
|
||
ofmatch->metadata = match->flow.metadata;
|
||
ofmatch->metadata_mask = ~match->wc.masks.metadata;
|
||
|
||
ofmatch->wildcards = htonl(wc);
|
||
}
|
||
|
||
/* Returns the "typical" length of a match for 'protocol', for use in
|
||
* estimating space to preallocate. */
|
||
int
|
||
ofputil_match_typical_len(enum ofputil_protocol protocol)
|
||
{
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
return sizeof(struct ofp10_match);
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
return NXM_TYPICAL_LEN;
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
return sizeof(struct ofp11_match);
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return NXM_TYPICAL_LEN;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Appends to 'b' an struct ofp11_match_header followed by a match that
|
||
* expresses 'match' properly for 'protocol', plus enough zero bytes to pad the
|
||
* data appended out to a multiple of 8. 'protocol' must be one that is usable
|
||
* in OpenFlow 1.1 or later.
|
||
*
|
||
* This function can cause 'b''s data to be reallocated.
|
||
*
|
||
* Returns the number of bytes appended to 'b', excluding the padding. Never
|
||
* returns zero. */
|
||
int
|
||
ofputil_put_ofp11_match(struct ofpbuf *b, const struct match *match,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
OVS_NOT_REACHED();
|
||
|
||
case OFPUTIL_P_OF11_STD: {
|
||
struct ofp11_match *om;
|
||
|
||
/* Make sure that no padding is needed. */
|
||
BUILD_ASSERT_DECL(sizeof *om % 8 == 0);
|
||
|
||
om = ofpbuf_put_uninit(b, sizeof *om);
|
||
ofputil_match_to_ofp11_match(match, om);
|
||
return sizeof *om;
|
||
}
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return oxm_put_match(b, match,
|
||
ofputil_protocol_to_ofp_version(protocol));
|
||
}
|
||
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
/* Given a 'dl_type' value in the format used in struct flow, returns the
|
||
* corresponding 'dl_type' value for use in an ofp10_match or ofp11_match
|
||
* structure. */
|
||
ovs_be16
|
||
ofputil_dl_type_to_openflow(ovs_be16 flow_dl_type)
|
||
{
|
||
return (flow_dl_type == htons(FLOW_DL_TYPE_NONE)
|
||
? htons(OFP_DL_TYPE_NOT_ETH_TYPE)
|
||
: flow_dl_type);
|
||
}
|
||
|
||
/* Given a 'dl_type' value in the format used in an ofp10_match or ofp11_match
|
||
* structure, returns the corresponding 'dl_type' value for use in struct
|
||
* flow. */
|
||
ovs_be16
|
||
ofputil_dl_type_from_openflow(ovs_be16 ofp_dl_type)
|
||
{
|
||
return (ofp_dl_type == htons(OFP_DL_TYPE_NOT_ETH_TYPE)
|
||
? htons(FLOW_DL_TYPE_NONE)
|
||
: ofp_dl_type);
|
||
}
|
||
|
||
/* Protocols. */
|
||
|
||
struct proto_abbrev {
|
||
enum ofputil_protocol protocol;
|
||
const char *name;
|
||
};
|
||
|
||
/* Most users really don't care about some of the differences between
|
||
* protocols. These abbreviations help with that. */
|
||
static const struct proto_abbrev proto_abbrevs[] = {
|
||
{ OFPUTIL_P_ANY, "any" },
|
||
{ OFPUTIL_P_OF10_STD_ANY, "OpenFlow10" },
|
||
{ OFPUTIL_P_OF10_NXM_ANY, "NXM" },
|
||
{ OFPUTIL_P_ANY_OXM, "OXM" },
|
||
};
|
||
#define N_PROTO_ABBREVS ARRAY_SIZE(proto_abbrevs)
|
||
|
||
enum ofputil_protocol ofputil_flow_dump_protocols[] = {
|
||
OFPUTIL_P_OF15_OXM,
|
||
OFPUTIL_P_OF14_OXM,
|
||
OFPUTIL_P_OF13_OXM,
|
||
OFPUTIL_P_OF12_OXM,
|
||
OFPUTIL_P_OF11_STD,
|
||
OFPUTIL_P_OF10_NXM,
|
||
OFPUTIL_P_OF10_STD,
|
||
};
|
||
size_t ofputil_n_flow_dump_protocols = ARRAY_SIZE(ofputil_flow_dump_protocols);
|
||
|
||
/* Returns the set of ofputil_protocols that are supported with the given
|
||
* OpenFlow 'version'. 'version' should normally be an 8-bit OpenFlow version
|
||
* identifier (e.g. 0x01 for OpenFlow 1.0, 0x02 for OpenFlow 1.1). Returns 0
|
||
* if 'version' is not supported or outside the valid range. */
|
||
enum ofputil_protocol
|
||
ofputil_protocols_from_ofp_version(enum ofp_version version)
|
||
{
|
||
switch (version) {
|
||
case OFP10_VERSION:
|
||
return OFPUTIL_P_OF10_STD_ANY | OFPUTIL_P_OF10_NXM_ANY;
|
||
case OFP11_VERSION:
|
||
return OFPUTIL_P_OF11_STD;
|
||
case OFP12_VERSION:
|
||
return OFPUTIL_P_OF12_OXM;
|
||
case OFP13_VERSION:
|
||
return OFPUTIL_P_OF13_OXM;
|
||
case OFP14_VERSION:
|
||
return OFPUTIL_P_OF14_OXM;
|
||
case OFP15_VERSION:
|
||
return OFPUTIL_P_OF15_OXM;
|
||
default:
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/* Returns the ofputil_protocol that is initially in effect on an OpenFlow
|
||
* connection that has negotiated the given 'version'. 'version' should
|
||
* normally be an 8-bit OpenFlow version identifier (e.g. 0x01 for OpenFlow
|
||
* 1.0, 0x02 for OpenFlow 1.1). Returns 0 if 'version' is not supported or
|
||
* outside the valid range. */
|
||
enum ofputil_protocol
|
||
ofputil_protocol_from_ofp_version(enum ofp_version version)
|
||
{
|
||
return rightmost_1bit(ofputil_protocols_from_ofp_version(version));
|
||
}
|
||
|
||
/* Returns the OpenFlow protocol version number (e.g. OFP10_VERSION,
|
||
* etc.) that corresponds to 'protocol'. */
|
||
enum ofp_version
|
||
ofputil_protocol_to_ofp_version(enum ofputil_protocol protocol)
|
||
{
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
return OFP10_VERSION;
|
||
case OFPUTIL_P_OF11_STD:
|
||
return OFP11_VERSION;
|
||
case OFPUTIL_P_OF12_OXM:
|
||
return OFP12_VERSION;
|
||
case OFPUTIL_P_OF13_OXM:
|
||
return OFP13_VERSION;
|
||
case OFPUTIL_P_OF14_OXM:
|
||
return OFP14_VERSION;
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return OFP15_VERSION;
|
||
}
|
||
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
/* Returns a bitmap of OpenFlow versions that are supported by at
|
||
* least one of the 'protocols'. */
|
||
uint32_t
|
||
ofputil_protocols_to_version_bitmap(enum ofputil_protocol protocols)
|
||
{
|
||
uint32_t bitmap = 0;
|
||
|
||
for (; protocols; protocols = zero_rightmost_1bit(protocols)) {
|
||
enum ofputil_protocol protocol = rightmost_1bit(protocols);
|
||
|
||
bitmap |= 1u << ofputil_protocol_to_ofp_version(protocol);
|
||
}
|
||
|
||
return bitmap;
|
||
}
|
||
|
||
/* Returns the set of protocols that are supported on top of the
|
||
* OpenFlow versions included in 'bitmap'. */
|
||
enum ofputil_protocol
|
||
ofputil_protocols_from_version_bitmap(uint32_t bitmap)
|
||
{
|
||
enum ofputil_protocol protocols = 0;
|
||
|
||
for (; bitmap; bitmap = zero_rightmost_1bit(bitmap)) {
|
||
enum ofp_version version = rightmost_1bit_idx(bitmap);
|
||
|
||
protocols |= ofputil_protocols_from_ofp_version(version);
|
||
}
|
||
|
||
return protocols;
|
||
}
|
||
|
||
/* Returns true if 'protocol' is a single OFPUTIL_P_* value, false
|
||
* otherwise. */
|
||
bool
|
||
ofputil_protocol_is_valid(enum ofputil_protocol protocol)
|
||
{
|
||
return protocol & OFPUTIL_P_ANY && is_pow2(protocol);
|
||
}
|
||
|
||
/* Returns the equivalent of 'protocol' with the Nicira flow_mod_table_id
|
||
* extension turned on or off if 'enable' is true or false, respectively.
|
||
*
|
||
* This extension is only useful for protocols whose "standard" version does
|
||
* not allow specific tables to be modified. In particular, this is true of
|
||
* OpenFlow 1.0. In later versions of OpenFlow, a flow_mod request always
|
||
* specifies a table ID and so there is no need for such an extension. When
|
||
* 'protocol' is such a protocol that doesn't need a flow_mod_table_id
|
||
* extension, this function just returns its 'protocol' argument unchanged
|
||
* regardless of the value of 'enable'. */
|
||
enum ofputil_protocol
|
||
ofputil_protocol_set_tid(enum ofputil_protocol protocol, bool enable)
|
||
{
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
return enable ? OFPUTIL_P_OF10_STD_TID : OFPUTIL_P_OF10_STD;
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
return enable ? OFPUTIL_P_OF10_NXM_TID : OFPUTIL_P_OF10_NXM;
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
return OFPUTIL_P_OF11_STD;
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
return OFPUTIL_P_OF12_OXM;
|
||
|
||
case OFPUTIL_P_OF13_OXM:
|
||
return OFPUTIL_P_OF13_OXM;
|
||
|
||
case OFPUTIL_P_OF14_OXM:
|
||
return OFPUTIL_P_OF14_OXM;
|
||
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return OFPUTIL_P_OF15_OXM;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Returns the "base" version of 'protocol'. That is, if 'protocol' includes
|
||
* some extension to a standard protocol version, the return value is the
|
||
* standard version of that protocol without any extension. If 'protocol' is a
|
||
* standard protocol version, returns 'protocol' unchanged. */
|
||
enum ofputil_protocol
|
||
ofputil_protocol_to_base(enum ofputil_protocol protocol)
|
||
{
|
||
return ofputil_protocol_set_tid(protocol, false);
|
||
}
|
||
|
||
/* Returns 'new_base' with any extensions taken from 'cur'. */
|
||
enum ofputil_protocol
|
||
ofputil_protocol_set_base(enum ofputil_protocol cur,
|
||
enum ofputil_protocol new_base)
|
||
{
|
||
bool tid = (cur & OFPUTIL_P_TID) != 0;
|
||
|
||
switch (new_base) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF10_STD, tid);
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF10_NXM, tid);
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF11_STD, tid);
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF12_OXM, tid);
|
||
|
||
case OFPUTIL_P_OF13_OXM:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF13_OXM, tid);
|
||
|
||
case OFPUTIL_P_OF14_OXM:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF14_OXM, tid);
|
||
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return ofputil_protocol_set_tid(OFPUTIL_P_OF15_OXM, tid);
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Returns a string form of 'protocol', if a simple form exists (that is, if
|
||
* 'protocol' is either a single protocol or it is a combination of protocols
|
||
* that have a single abbreviation). Otherwise, returns NULL. */
|
||
const char *
|
||
ofputil_protocol_to_string(enum ofputil_protocol protocol)
|
||
{
|
||
const struct proto_abbrev *p;
|
||
|
||
/* Use a "switch" statement for single-bit names so that we get a compiler
|
||
* warning if we forget any. */
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_NXM:
|
||
return "NXM-table_id";
|
||
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
return "NXM+table_id";
|
||
|
||
case OFPUTIL_P_OF10_STD:
|
||
return "OpenFlow10-table_id";
|
||
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
return "OpenFlow10+table_id";
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
return "OpenFlow11";
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
return "OXM-OpenFlow12";
|
||
|
||
case OFPUTIL_P_OF13_OXM:
|
||
return "OXM-OpenFlow13";
|
||
|
||
case OFPUTIL_P_OF14_OXM:
|
||
return "OXM-OpenFlow14";
|
||
|
||
case OFPUTIL_P_OF15_OXM:
|
||
return "OXM-OpenFlow15";
|
||
}
|
||
|
||
/* Check abbreviations. */
|
||
for (p = proto_abbrevs; p < &proto_abbrevs[N_PROTO_ABBREVS]; p++) {
|
||
if (protocol == p->protocol) {
|
||
return p->name;
|
||
}
|
||
}
|
||
|
||
return NULL;
|
||
}
|
||
|
||
/* Returns a string that represents 'protocols'. The return value might be a
|
||
* comma-separated list if 'protocols' doesn't have a simple name. The return
|
||
* value is "none" if 'protocols' is 0.
|
||
*
|
||
* The caller must free the returned string (with free()). */
|
||
char *
|
||
ofputil_protocols_to_string(enum ofputil_protocol protocols)
|
||
{
|
||
struct ds s;
|
||
|
||
ovs_assert(!(protocols & ~OFPUTIL_P_ANY));
|
||
if (protocols == 0) {
|
||
return xstrdup("none");
|
||
}
|
||
|
||
ds_init(&s);
|
||
while (protocols) {
|
||
const struct proto_abbrev *p;
|
||
int i;
|
||
|
||
if (s.length) {
|
||
ds_put_char(&s, ',');
|
||
}
|
||
|
||
for (p = proto_abbrevs; p < &proto_abbrevs[N_PROTO_ABBREVS]; p++) {
|
||
if ((protocols & p->protocol) == p->protocol) {
|
||
ds_put_cstr(&s, p->name);
|
||
protocols &= ~p->protocol;
|
||
goto match;
|
||
}
|
||
}
|
||
|
||
for (i = 0; i < CHAR_BIT * sizeof(enum ofputil_protocol); i++) {
|
||
enum ofputil_protocol bit = 1u << i;
|
||
|
||
if (protocols & bit) {
|
||
ds_put_cstr(&s, ofputil_protocol_to_string(bit));
|
||
protocols &= ~bit;
|
||
goto match;
|
||
}
|
||
}
|
||
OVS_NOT_REACHED();
|
||
|
||
match: ;
|
||
}
|
||
return ds_steal_cstr(&s);
|
||
}
|
||
|
||
static enum ofputil_protocol
|
||
ofputil_protocol_from_string__(const char *s, size_t n)
|
||
{
|
||
const struct proto_abbrev *p;
|
||
int i;
|
||
|
||
for (i = 0; i < CHAR_BIT * sizeof(enum ofputil_protocol); i++) {
|
||
enum ofputil_protocol bit = 1u << i;
|
||
const char *name = ofputil_protocol_to_string(bit);
|
||
|
||
if (name && n == strlen(name) && !strncasecmp(s, name, n)) {
|
||
return bit;
|
||
}
|
||
}
|
||
|
||
for (p = proto_abbrevs; p < &proto_abbrevs[N_PROTO_ABBREVS]; p++) {
|
||
if (n == strlen(p->name) && !strncasecmp(s, p->name, n)) {
|
||
return p->protocol;
|
||
}
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Returns the nonempty set of protocols represented by 's', which can be a
|
||
* single protocol name or abbreviation or a comma-separated list of them.
|
||
*
|
||
* Aborts the program with an error message if 's' is invalid. */
|
||
enum ofputil_protocol
|
||
ofputil_protocols_from_string(const char *s)
|
||
{
|
||
const char *orig_s = s;
|
||
enum ofputil_protocol protocols;
|
||
|
||
protocols = 0;
|
||
while (*s) {
|
||
enum ofputil_protocol p;
|
||
size_t n;
|
||
|
||
n = strcspn(s, ",");
|
||
if (n == 0) {
|
||
s++;
|
||
continue;
|
||
}
|
||
|
||
p = ofputil_protocol_from_string__(s, n);
|
||
if (!p) {
|
||
ovs_fatal(0, "%.*s: unknown flow protocol", (int) n, s);
|
||
}
|
||
protocols |= p;
|
||
|
||
s += n;
|
||
}
|
||
|
||
if (!protocols) {
|
||
ovs_fatal(0, "%s: no flow protocol specified", orig_s);
|
||
}
|
||
return protocols;
|
||
}
|
||
|
||
enum ofp_version
|
||
ofputil_version_from_string(const char *s)
|
||
{
|
||
if (!strcasecmp(s, "OpenFlow10")) {
|
||
return OFP10_VERSION;
|
||
}
|
||
if (!strcasecmp(s, "OpenFlow11")) {
|
||
return OFP11_VERSION;
|
||
}
|
||
if (!strcasecmp(s, "OpenFlow12")) {
|
||
return OFP12_VERSION;
|
||
}
|
||
if (!strcasecmp(s, "OpenFlow13")) {
|
||
return OFP13_VERSION;
|
||
}
|
||
if (!strcasecmp(s, "OpenFlow14")) {
|
||
return OFP14_VERSION;
|
||
}
|
||
if (!strcasecmp(s, "OpenFlow15")) {
|
||
return OFP15_VERSION;
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
static bool
|
||
is_delimiter(unsigned char c)
|
||
{
|
||
return isspace(c) || c == ',';
|
||
}
|
||
|
||
uint32_t
|
||
ofputil_versions_from_string(const char *s)
|
||
{
|
||
size_t i = 0;
|
||
uint32_t bitmap = 0;
|
||
|
||
while (s[i]) {
|
||
size_t j;
|
||
int version;
|
||
char *key;
|
||
|
||
if (is_delimiter(s[i])) {
|
||
i++;
|
||
continue;
|
||
}
|
||
j = 0;
|
||
while (s[i + j] && !is_delimiter(s[i + j])) {
|
||
j++;
|
||
}
|
||
key = xmemdup0(s + i, j);
|
||
version = ofputil_version_from_string(key);
|
||
if (!version) {
|
||
VLOG_FATAL("Unknown OpenFlow version: \"%s\"", key);
|
||
}
|
||
free(key);
|
||
bitmap |= 1u << version;
|
||
i += j;
|
||
}
|
||
|
||
return bitmap;
|
||
}
|
||
|
||
uint32_t
|
||
ofputil_versions_from_strings(char ** const s, size_t count)
|
||
{
|
||
uint32_t bitmap = 0;
|
||
|
||
while (count--) {
|
||
int version = ofputil_version_from_string(s[count]);
|
||
if (!version) {
|
||
VLOG_WARN("Unknown OpenFlow version: \"%s\"", s[count]);
|
||
} else {
|
||
bitmap |= 1u << version;
|
||
}
|
||
}
|
||
|
||
return bitmap;
|
||
}
|
||
|
||
const char *
|
||
ofputil_version_to_string(enum ofp_version ofp_version)
|
||
{
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
return "OpenFlow10";
|
||
case OFP11_VERSION:
|
||
return "OpenFlow11";
|
||
case OFP12_VERSION:
|
||
return "OpenFlow12";
|
||
case OFP13_VERSION:
|
||
return "OpenFlow13";
|
||
case OFP14_VERSION:
|
||
return "OpenFlow14";
|
||
case OFP15_VERSION:
|
||
return "OpenFlow15";
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
bool
|
||
ofputil_packet_in_format_is_valid(enum nx_packet_in_format packet_in_format)
|
||
{
|
||
switch (packet_in_format) {
|
||
case NXPIF_OPENFLOW10:
|
||
case NXPIF_NXM:
|
||
return true;
|
||
}
|
||
|
||
return false;
|
||
}
|
||
|
||
const char *
|
||
ofputil_packet_in_format_to_string(enum nx_packet_in_format packet_in_format)
|
||
{
|
||
switch (packet_in_format) {
|
||
case NXPIF_OPENFLOW10:
|
||
return "openflow10";
|
||
case NXPIF_NXM:
|
||
return "nxm";
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
int
|
||
ofputil_packet_in_format_from_string(const char *s)
|
||
{
|
||
return (!strcmp(s, "openflow10") ? NXPIF_OPENFLOW10
|
||
: !strcmp(s, "nxm") ? NXPIF_NXM
|
||
: -1);
|
||
}
|
||
|
||
void
|
||
ofputil_format_version(struct ds *msg, enum ofp_version version)
|
||
{
|
||
ds_put_format(msg, "0x%02x", version);
|
||
}
|
||
|
||
void
|
||
ofputil_format_version_name(struct ds *msg, enum ofp_version version)
|
||
{
|
||
ds_put_cstr(msg, ofputil_version_to_string(version));
|
||
}
|
||
|
||
static void
|
||
ofputil_format_version_bitmap__(struct ds *msg, uint32_t bitmap,
|
||
void (*format_version)(struct ds *msg,
|
||
enum ofp_version))
|
||
{
|
||
while (bitmap) {
|
||
format_version(msg, raw_ctz(bitmap));
|
||
bitmap = zero_rightmost_1bit(bitmap);
|
||
if (bitmap) {
|
||
ds_put_cstr(msg, ", ");
|
||
}
|
||
}
|
||
}
|
||
|
||
void
|
||
ofputil_format_version_bitmap(struct ds *msg, uint32_t bitmap)
|
||
{
|
||
ofputil_format_version_bitmap__(msg, bitmap, ofputil_format_version);
|
||
}
|
||
|
||
void
|
||
ofputil_format_version_bitmap_names(struct ds *msg, uint32_t bitmap)
|
||
{
|
||
ofputil_format_version_bitmap__(msg, bitmap, ofputil_format_version_name);
|
||
}
|
||
|
||
static bool
|
||
ofputil_decode_hello_bitmap(const struct ofp_hello_elem_header *oheh,
|
||
uint32_t *allowed_versionsp)
|
||
{
|
||
uint16_t bitmap_len = ntohs(oheh->length) - sizeof *oheh;
|
||
const ovs_be32 *bitmap = ALIGNED_CAST(const ovs_be32 *, oheh + 1);
|
||
uint32_t allowed_versions;
|
||
|
||
if (!bitmap_len || bitmap_len % sizeof *bitmap) {
|
||
return false;
|
||
}
|
||
|
||
/* Only use the first 32-bit element of the bitmap as that is all the
|
||
* current implementation supports. Subsequent elements are ignored which
|
||
* should have no effect on session negotiation until Open vSwtich supports
|
||
* wire-protocol versions greater than 31.
|
||
*/
|
||
allowed_versions = ntohl(bitmap[0]);
|
||
|
||
if (allowed_versions & 1) {
|
||
/* There's no OpenFlow version 0. */
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "peer claims to support invalid OpenFlow "
|
||
"version 0x00");
|
||
allowed_versions &= ~1u;
|
||
}
|
||
|
||
if (!allowed_versions) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "peer does not support any OpenFlow "
|
||
"version (between 0x01 and 0x1f)");
|
||
return false;
|
||
}
|
||
|
||
*allowed_versionsp = allowed_versions;
|
||
return true;
|
||
}
|
||
|
||
static uint32_t
|
||
version_bitmap_from_version(uint8_t ofp_version)
|
||
{
|
||
return ((ofp_version < 32 ? 1u << ofp_version : 0) - 1) << 1;
|
||
}
|
||
|
||
/* Decodes OpenFlow OFPT_HELLO message 'oh', storing into '*allowed_versions'
|
||
* the set of OpenFlow versions for which 'oh' announces support.
|
||
*
|
||
* Because of how OpenFlow defines OFPT_HELLO messages, this function is always
|
||
* successful, and thus '*allowed_versions' is always initialized. However, it
|
||
* returns false if 'oh' contains some data that could not be fully understood,
|
||
* true if 'oh' was completely parsed. */
|
||
bool
|
||
ofputil_decode_hello(const struct ofp_header *oh, uint32_t *allowed_versions)
|
||
{
|
||
struct ofpbuf msg;
|
||
bool ok = true;
|
||
|
||
ofpbuf_use_const(&msg, oh, ntohs(oh->length));
|
||
ofpbuf_pull(&msg, sizeof *oh);
|
||
|
||
*allowed_versions = version_bitmap_from_version(oh->version);
|
||
while (ofpbuf_size(&msg)) {
|
||
const struct ofp_hello_elem_header *oheh;
|
||
unsigned int len;
|
||
|
||
if (ofpbuf_size(&msg) < sizeof *oheh) {
|
||
return false;
|
||
}
|
||
|
||
oheh = ofpbuf_data(&msg);
|
||
len = ntohs(oheh->length);
|
||
if (len < sizeof *oheh || !ofpbuf_try_pull(&msg, ROUND_UP(len, 8))) {
|
||
return false;
|
||
}
|
||
|
||
if (oheh->type != htons(OFPHET_VERSIONBITMAP)
|
||
|| !ofputil_decode_hello_bitmap(oheh, allowed_versions)) {
|
||
ok = false;
|
||
}
|
||
}
|
||
|
||
return ok;
|
||
}
|
||
|
||
/* Returns true if 'allowed_versions' needs to be accompanied by a version
|
||
* bitmap to be correctly expressed in an OFPT_HELLO message. */
|
||
static bool
|
||
should_send_version_bitmap(uint32_t allowed_versions)
|
||
{
|
||
return !is_pow2((allowed_versions >> 1) + 1);
|
||
}
|
||
|
||
/* Create an OFPT_HELLO message that expresses support for the OpenFlow
|
||
* versions in the 'allowed_versions' bitmaps and returns the message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_hello(uint32_t allowed_versions)
|
||
{
|
||
enum ofp_version ofp_version;
|
||
struct ofpbuf *msg;
|
||
|
||
ofp_version = leftmost_1bit_idx(allowed_versions);
|
||
msg = ofpraw_alloc(OFPRAW_OFPT_HELLO, ofp_version, 0);
|
||
|
||
if (should_send_version_bitmap(allowed_versions)) {
|
||
struct ofp_hello_elem_header *oheh;
|
||
uint16_t map_len;
|
||
|
||
map_len = sizeof allowed_versions;
|
||
oheh = ofpbuf_put_zeros(msg, ROUND_UP(map_len + sizeof *oheh, 8));
|
||
oheh->type = htons(OFPHET_VERSIONBITMAP);
|
||
oheh->length = htons(map_len + sizeof *oheh);
|
||
*ALIGNED_CAST(ovs_be32 *, oheh + 1) = htonl(allowed_versions);
|
||
|
||
ofpmsg_update_length(msg);
|
||
}
|
||
|
||
return msg;
|
||
}
|
||
|
||
/* Returns an OpenFlow message that, sent on an OpenFlow connection whose
|
||
* protocol is 'current', at least partly transitions the protocol to 'want'.
|
||
* Stores in '*next' the protocol that will be in effect on the OpenFlow
|
||
* connection if the switch processes the returned message correctly. (If
|
||
* '*next != want' then the caller will have to iterate.)
|
||
*
|
||
* If 'current == want', or if it is not possible to transition from 'current'
|
||
* to 'want' (because, for example, 'current' and 'want' use different OpenFlow
|
||
* protocol versions), returns NULL and stores 'current' in '*next'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_set_protocol(enum ofputil_protocol current,
|
||
enum ofputil_protocol want,
|
||
enum ofputil_protocol *next)
|
||
{
|
||
enum ofp_version cur_version, want_version;
|
||
enum ofputil_protocol cur_base, want_base;
|
||
bool cur_tid, want_tid;
|
||
|
||
cur_version = ofputil_protocol_to_ofp_version(current);
|
||
want_version = ofputil_protocol_to_ofp_version(want);
|
||
if (cur_version != want_version) {
|
||
*next = current;
|
||
return NULL;
|
||
}
|
||
|
||
cur_base = ofputil_protocol_to_base(current);
|
||
want_base = ofputil_protocol_to_base(want);
|
||
if (cur_base != want_base) {
|
||
*next = ofputil_protocol_set_base(current, want_base);
|
||
|
||
switch (want_base) {
|
||
case OFPUTIL_P_OF10_NXM:
|
||
return ofputil_encode_nx_set_flow_format(NXFF_NXM);
|
||
|
||
case OFPUTIL_P_OF10_STD:
|
||
return ofputil_encode_nx_set_flow_format(NXFF_OPENFLOW10);
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM:
|
||
/* There is only one variant of each OpenFlow 1.1+ protocol, and we
|
||
* verified above that we're not trying to change versions. */
|
||
OVS_NOT_REACHED();
|
||
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
cur_tid = (current & OFPUTIL_P_TID) != 0;
|
||
want_tid = (want & OFPUTIL_P_TID) != 0;
|
||
if (cur_tid != want_tid) {
|
||
*next = ofputil_protocol_set_tid(current, want_tid);
|
||
return ofputil_make_flow_mod_table_id(want_tid);
|
||
}
|
||
|
||
ovs_assert(current == want);
|
||
|
||
*next = current;
|
||
return NULL;
|
||
}
|
||
|
||
/* Returns an NXT_SET_FLOW_FORMAT message that can be used to set the flow
|
||
* format to 'nxff'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_nx_set_flow_format(enum nx_flow_format nxff)
|
||
{
|
||
struct nx_set_flow_format *sff;
|
||
struct ofpbuf *msg;
|
||
|
||
ovs_assert(ofputil_nx_flow_format_is_valid(nxff));
|
||
|
||
msg = ofpraw_alloc(OFPRAW_NXT_SET_FLOW_FORMAT, OFP10_VERSION, 0);
|
||
sff = ofpbuf_put_zeros(msg, sizeof *sff);
|
||
sff->format = htonl(nxff);
|
||
|
||
return msg;
|
||
}
|
||
|
||
/* Returns the base protocol if 'flow_format' is a valid NXFF_* value, false
|
||
* otherwise. */
|
||
enum ofputil_protocol
|
||
ofputil_nx_flow_format_to_protocol(enum nx_flow_format flow_format)
|
||
{
|
||
switch (flow_format) {
|
||
case NXFF_OPENFLOW10:
|
||
return OFPUTIL_P_OF10_STD;
|
||
|
||
case NXFF_NXM:
|
||
return OFPUTIL_P_OF10_NXM;
|
||
|
||
default:
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/* Returns true if 'flow_format' is a valid NXFF_* value, false otherwise. */
|
||
bool
|
||
ofputil_nx_flow_format_is_valid(enum nx_flow_format flow_format)
|
||
{
|
||
return ofputil_nx_flow_format_to_protocol(flow_format) != 0;
|
||
}
|
||
|
||
/* Returns a string version of 'flow_format', which must be a valid NXFF_*
|
||
* value. */
|
||
const char *
|
||
ofputil_nx_flow_format_to_string(enum nx_flow_format flow_format)
|
||
{
|
||
switch (flow_format) {
|
||
case NXFF_OPENFLOW10:
|
||
return "openflow10";
|
||
case NXFF_NXM:
|
||
return "nxm";
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_make_set_packet_in_format(enum ofp_version ofp_version,
|
||
enum nx_packet_in_format packet_in_format)
|
||
{
|
||
struct nx_set_packet_in_format *spif;
|
||
struct ofpbuf *msg;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_NXT_SET_PACKET_IN_FORMAT, ofp_version, 0);
|
||
spif = ofpbuf_put_zeros(msg, sizeof *spif);
|
||
spif->format = htonl(packet_in_format);
|
||
|
||
return msg;
|
||
}
|
||
|
||
/* Returns an OpenFlow message that can be used to turn the flow_mod_table_id
|
||
* extension on or off (according to 'flow_mod_table_id'). */
|
||
struct ofpbuf *
|
||
ofputil_make_flow_mod_table_id(bool flow_mod_table_id)
|
||
{
|
||
struct nx_flow_mod_table_id *nfmti;
|
||
struct ofpbuf *msg;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_NXT_FLOW_MOD_TABLE_ID, OFP10_VERSION, 0);
|
||
nfmti = ofpbuf_put_zeros(msg, sizeof *nfmti);
|
||
nfmti->set = flow_mod_table_id;
|
||
return msg;
|
||
}
|
||
|
||
struct ofputil_flow_mod_flag {
|
||
uint16_t raw_flag;
|
||
enum ofp_version min_version, max_version;
|
||
enum ofputil_flow_mod_flags flag;
|
||
};
|
||
|
||
static const struct ofputil_flow_mod_flag ofputil_flow_mod_flags[] = {
|
||
{ OFPFF_SEND_FLOW_REM, OFP10_VERSION, 0, OFPUTIL_FF_SEND_FLOW_REM },
|
||
{ OFPFF_CHECK_OVERLAP, OFP10_VERSION, 0, OFPUTIL_FF_CHECK_OVERLAP },
|
||
{ OFPFF10_EMERG, OFP10_VERSION, OFP10_VERSION,
|
||
OFPUTIL_FF_EMERG },
|
||
{ OFPFF12_RESET_COUNTS, OFP12_VERSION, 0, OFPUTIL_FF_RESET_COUNTS },
|
||
{ OFPFF13_NO_PKT_COUNTS, OFP13_VERSION, 0, OFPUTIL_FF_NO_PKT_COUNTS },
|
||
{ OFPFF13_NO_BYT_COUNTS, OFP13_VERSION, 0, OFPUTIL_FF_NO_BYT_COUNTS },
|
||
{ 0, 0, 0, 0 },
|
||
};
|
||
|
||
static enum ofperr
|
||
ofputil_decode_flow_mod_flags(ovs_be16 raw_flags_,
|
||
enum ofp_flow_mod_command command,
|
||
enum ofp_version version,
|
||
enum ofputil_flow_mod_flags *flagsp)
|
||
{
|
||
uint16_t raw_flags = ntohs(raw_flags_);
|
||
const struct ofputil_flow_mod_flag *f;
|
||
|
||
*flagsp = 0;
|
||
for (f = ofputil_flow_mod_flags; f->raw_flag; f++) {
|
||
if (raw_flags & f->raw_flag
|
||
&& version >= f->min_version
|
||
&& (!f->max_version || version <= f->max_version)) {
|
||
raw_flags &= ~f->raw_flag;
|
||
*flagsp |= f->flag;
|
||
}
|
||
}
|
||
|
||
/* In OF1.0 and OF1.1, "add" always resets counters, and other commands
|
||
* never do.
|
||
*
|
||
* In OF1.2 and later, OFPFF12_RESET_COUNTS controls whether each command
|
||
* resets counters. */
|
||
if ((version == OFP10_VERSION || version == OFP11_VERSION)
|
||
&& command == OFPFC_ADD) {
|
||
*flagsp |= OFPUTIL_FF_RESET_COUNTS;
|
||
}
|
||
|
||
return raw_flags ? OFPERR_OFPFMFC_BAD_FLAGS : 0;
|
||
}
|
||
|
||
static ovs_be16
|
||
ofputil_encode_flow_mod_flags(enum ofputil_flow_mod_flags flags,
|
||
enum ofp_version version)
|
||
{
|
||
const struct ofputil_flow_mod_flag *f;
|
||
uint16_t raw_flags;
|
||
|
||
raw_flags = 0;
|
||
for (f = ofputil_flow_mod_flags; f->raw_flag; f++) {
|
||
if (f->flag & flags
|
||
&& version >= f->min_version
|
||
&& (!f->max_version || version <= f->max_version)) {
|
||
raw_flags |= f->raw_flag;
|
||
}
|
||
}
|
||
|
||
return htons(raw_flags);
|
||
}
|
||
|
||
/* Converts an OFPT_FLOW_MOD or NXT_FLOW_MOD message 'oh' into an abstract
|
||
* flow_mod in 'fm'. Returns 0 if successful, otherwise an OpenFlow error
|
||
* code.
|
||
*
|
||
* Uses 'ofpacts' to store the abstract OFPACT_* version of 'oh''s actions.
|
||
* The caller must initialize 'ofpacts' and retains ownership of it.
|
||
* 'fm->ofpacts' will point into the 'ofpacts' buffer.
|
||
*
|
||
* Does not validate the flow_mod actions. The caller should do that, with
|
||
* ofpacts_check(). */
|
||
enum ofperr
|
||
ofputil_decode_flow_mod(struct ofputil_flow_mod *fm,
|
||
const struct ofp_header *oh,
|
||
enum ofputil_protocol protocol,
|
||
struct ofpbuf *ofpacts,
|
||
ofp_port_t max_port, uint8_t max_table)
|
||
{
|
||
ovs_be16 raw_flags;
|
||
enum ofperr error;
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
|
||
/* Ignored for non-delete actions */
|
||
fm->delete_reason = OFPRR_DELETE;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
if (raw == OFPRAW_OFPT11_FLOW_MOD) {
|
||
/* Standard OpenFlow 1.1+ flow_mod. */
|
||
const struct ofp11_flow_mod *ofm;
|
||
|
||
ofm = ofpbuf_pull(&b, sizeof *ofm);
|
||
|
||
error = ofputil_pull_ofp11_match(&b, &fm->match, NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
/* Translate the message. */
|
||
fm->priority = ntohs(ofm->priority);
|
||
if (ofm->command == OFPFC_ADD
|
||
|| (oh->version == OFP11_VERSION
|
||
&& (ofm->command == OFPFC_MODIFY ||
|
||
ofm->command == OFPFC_MODIFY_STRICT)
|
||
&& ofm->cookie_mask == htonll(0))) {
|
||
/* In OpenFlow 1.1 only, a "modify" or "modify-strict" that does
|
||
* not match on the cookie is treated as an "add" if there is no
|
||
* match. */
|
||
fm->cookie = htonll(0);
|
||
fm->cookie_mask = htonll(0);
|
||
fm->new_cookie = ofm->cookie;
|
||
} else {
|
||
fm->cookie = ofm->cookie;
|
||
fm->cookie_mask = ofm->cookie_mask;
|
||
fm->new_cookie = OVS_BE64_MAX;
|
||
}
|
||
fm->modify_cookie = false;
|
||
fm->command = ofm->command;
|
||
|
||
/* Get table ID.
|
||
*
|
||
* OF1.1 entirely forbids table_id == OFPTT_ALL.
|
||
* OF1.2+ allows table_id == OFPTT_ALL only for deletes. */
|
||
fm->table_id = ofm->table_id;
|
||
if (fm->table_id == OFPTT_ALL
|
||
&& (oh->version == OFP11_VERSION
|
||
|| (ofm->command != OFPFC_DELETE &&
|
||
ofm->command != OFPFC_DELETE_STRICT))) {
|
||
return OFPERR_OFPFMFC_BAD_TABLE_ID;
|
||
}
|
||
|
||
fm->idle_timeout = ntohs(ofm->idle_timeout);
|
||
fm->hard_timeout = ntohs(ofm->hard_timeout);
|
||
fm->buffer_id = ntohl(ofm->buffer_id);
|
||
error = ofputil_port_from_ofp11(ofm->out_port, &fm->out_port);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
fm->out_group = (ofm->command == OFPFC_DELETE ||
|
||
ofm->command == OFPFC_DELETE_STRICT
|
||
? ntohl(ofm->out_group)
|
||
: OFPG11_ANY);
|
||
raw_flags = ofm->flags;
|
||
} else {
|
||
uint16_t command;
|
||
|
||
if (raw == OFPRAW_OFPT10_FLOW_MOD) {
|
||
/* Standard OpenFlow 1.0 flow_mod. */
|
||
const struct ofp10_flow_mod *ofm;
|
||
|
||
/* Get the ofp10_flow_mod. */
|
||
ofm = ofpbuf_pull(&b, sizeof *ofm);
|
||
|
||
/* Translate the rule. */
|
||
ofputil_match_from_ofp10_match(&ofm->match, &fm->match);
|
||
ofputil_normalize_match(&fm->match);
|
||
|
||
/* OpenFlow 1.0 says that exact-match rules have to have the
|
||
* highest possible priority. */
|
||
fm->priority = (ofm->match.wildcards & htonl(OFPFW10_ALL)
|
||
? ntohs(ofm->priority)
|
||
: UINT16_MAX);
|
||
|
||
/* Translate the message. */
|
||
command = ntohs(ofm->command);
|
||
fm->cookie = htonll(0);
|
||
fm->cookie_mask = htonll(0);
|
||
fm->new_cookie = ofm->cookie;
|
||
fm->idle_timeout = ntohs(ofm->idle_timeout);
|
||
fm->hard_timeout = ntohs(ofm->hard_timeout);
|
||
fm->buffer_id = ntohl(ofm->buffer_id);
|
||
fm->out_port = u16_to_ofp(ntohs(ofm->out_port));
|
||
fm->out_group = OFPG11_ANY;
|
||
raw_flags = ofm->flags;
|
||
} else if (raw == OFPRAW_NXT_FLOW_MOD) {
|
||
/* Nicira extended flow_mod. */
|
||
const struct nx_flow_mod *nfm;
|
||
|
||
/* Dissect the message. */
|
||
nfm = ofpbuf_pull(&b, sizeof *nfm);
|
||
error = nx_pull_match(&b, ntohs(nfm->match_len),
|
||
&fm->match, &fm->cookie, &fm->cookie_mask);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
/* Translate the message. */
|
||
command = ntohs(nfm->command);
|
||
if ((command & 0xff) == OFPFC_ADD && fm->cookie_mask) {
|
||
/* Flow additions may only set a new cookie, not match an
|
||
* existing cookie. */
|
||
return OFPERR_NXBRC_NXM_INVALID;
|
||
}
|
||
fm->priority = ntohs(nfm->priority);
|
||
fm->new_cookie = nfm->cookie;
|
||
fm->idle_timeout = ntohs(nfm->idle_timeout);
|
||
fm->hard_timeout = ntohs(nfm->hard_timeout);
|
||
fm->buffer_id = ntohl(nfm->buffer_id);
|
||
fm->out_port = u16_to_ofp(ntohs(nfm->out_port));
|
||
fm->out_group = OFPG11_ANY;
|
||
raw_flags = nfm->flags;
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
fm->modify_cookie = fm->new_cookie != OVS_BE64_MAX;
|
||
if (protocol & OFPUTIL_P_TID) {
|
||
fm->command = command & 0xff;
|
||
fm->table_id = command >> 8;
|
||
} else {
|
||
fm->command = command;
|
||
fm->table_id = 0xff;
|
||
}
|
||
}
|
||
|
||
error = ofpacts_pull_openflow_instructions(&b, ofpbuf_size(&b),
|
||
oh->version, ofpacts);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
fm->ofpacts = ofpbuf_data(ofpacts);
|
||
fm->ofpacts_len = ofpbuf_size(ofpacts);
|
||
|
||
error = ofputil_decode_flow_mod_flags(raw_flags, fm->command,
|
||
oh->version, &fm->flags);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (fm->flags & OFPUTIL_FF_EMERG) {
|
||
/* We do not support the OpenFlow 1.0 emergency flow cache, which
|
||
* is not required in OpenFlow 1.0.1 and removed from OpenFlow 1.1.
|
||
*
|
||
* OpenFlow 1.0 specifies the error code to use when idle_timeout
|
||
* or hard_timeout is nonzero. Otherwise, there is no good error
|
||
* code, so just state that the flow table is full. */
|
||
return (fm->hard_timeout || fm->idle_timeout
|
||
? OFPERR_OFPFMFC_BAD_EMERG_TIMEOUT
|
||
: OFPERR_OFPFMFC_TABLE_FULL);
|
||
}
|
||
|
||
return ofpacts_check_consistency(fm->ofpacts, fm->ofpacts_len,
|
||
&fm->match.flow, max_port,
|
||
fm->table_id, max_table, protocol);
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_bands(struct ofpbuf *msg, size_t len, uint16_t *n_bands,
|
||
struct ofpbuf *bands)
|
||
{
|
||
const struct ofp13_meter_band_header *ombh;
|
||
struct ofputil_meter_band *mb;
|
||
uint16_t n = 0;
|
||
|
||
ombh = ofpbuf_try_pull(msg, len);
|
||
if (!ombh) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
while (len >= sizeof (struct ofp13_meter_band_drop)) {
|
||
size_t ombh_len = ntohs(ombh->len);
|
||
/* All supported band types have the same length. */
|
||
if (ombh_len != sizeof (struct ofp13_meter_band_drop)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
mb = ofpbuf_put_uninit(bands, sizeof *mb);
|
||
mb->type = ntohs(ombh->type);
|
||
if (mb->type != OFPMBT13_DROP && mb->type != OFPMBT13_DSCP_REMARK) {
|
||
return OFPERR_OFPMMFC_BAD_BAND;
|
||
}
|
||
mb->rate = ntohl(ombh->rate);
|
||
mb->burst_size = ntohl(ombh->burst_size);
|
||
mb->prec_level = (mb->type == OFPMBT13_DSCP_REMARK) ?
|
||
((struct ofp13_meter_band_dscp_remark *)ombh)->prec_level : 0;
|
||
n++;
|
||
len -= ombh_len;
|
||
ombh = ALIGNED_CAST(struct ofp13_meter_band_header *,
|
||
(char *) ombh + ombh_len);
|
||
}
|
||
if (len) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
*n_bands = n;
|
||
return 0;
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_meter_mod(const struct ofp_header *oh,
|
||
struct ofputil_meter_mod *mm,
|
||
struct ofpbuf *bands)
|
||
{
|
||
const struct ofp13_meter_mod *omm;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
ofpraw_pull_assert(&b);
|
||
omm = ofpbuf_pull(&b, sizeof *omm);
|
||
|
||
/* Translate the message. */
|
||
mm->command = ntohs(omm->command);
|
||
if (mm->command != OFPMC13_ADD &&
|
||
mm->command != OFPMC13_MODIFY &&
|
||
mm->command != OFPMC13_DELETE) {
|
||
return OFPERR_OFPMMFC_BAD_COMMAND;
|
||
}
|
||
mm->meter.meter_id = ntohl(omm->meter_id);
|
||
|
||
if (mm->command == OFPMC13_DELETE) {
|
||
mm->meter.flags = 0;
|
||
mm->meter.n_bands = 0;
|
||
mm->meter.bands = NULL;
|
||
} else {
|
||
enum ofperr error;
|
||
|
||
mm->meter.flags = ntohs(omm->flags);
|
||
if (mm->meter.flags & OFPMF13_KBPS &&
|
||
mm->meter.flags & OFPMF13_PKTPS) {
|
||
return OFPERR_OFPMMFC_BAD_FLAGS;
|
||
}
|
||
mm->meter.bands = ofpbuf_data(bands);
|
||
|
||
error = ofputil_pull_bands(&b, ofpbuf_size(&b), &mm->meter.n_bands, bands);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
void
|
||
ofputil_decode_meter_request(const struct ofp_header *oh, uint32_t *meter_id)
|
||
{
|
||
const struct ofp13_meter_multipart_request *omr = ofpmsg_body(oh);
|
||
*meter_id = ntohl(omr->meter_id);
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_meter_request(enum ofp_version ofp_version,
|
||
enum ofputil_meter_request_type type,
|
||
uint32_t meter_id)
|
||
{
|
||
struct ofpbuf *msg;
|
||
|
||
enum ofpraw raw;
|
||
|
||
switch (type) {
|
||
case OFPUTIL_METER_CONFIG:
|
||
raw = OFPRAW_OFPST13_METER_CONFIG_REQUEST;
|
||
break;
|
||
case OFPUTIL_METER_STATS:
|
||
raw = OFPRAW_OFPST13_METER_REQUEST;
|
||
break;
|
||
default:
|
||
case OFPUTIL_METER_FEATURES:
|
||
raw = OFPRAW_OFPST13_METER_FEATURES_REQUEST;
|
||
break;
|
||
}
|
||
|
||
msg = ofpraw_alloc(raw, ofp_version, 0);
|
||
|
||
if (type != OFPUTIL_METER_FEATURES) {
|
||
struct ofp13_meter_multipart_request *omr;
|
||
omr = ofpbuf_put_zeros(msg, sizeof *omr);
|
||
omr->meter_id = htonl(meter_id);
|
||
}
|
||
return msg;
|
||
}
|
||
|
||
static void
|
||
ofputil_put_bands(uint16_t n_bands, const struct ofputil_meter_band *mb,
|
||
struct ofpbuf *msg)
|
||
{
|
||
uint16_t n = 0;
|
||
|
||
for (n = 0; n < n_bands; ++n) {
|
||
/* Currently all band types have same size. */
|
||
struct ofp13_meter_band_dscp_remark *ombh;
|
||
size_t ombh_len = sizeof *ombh;
|
||
|
||
ombh = ofpbuf_put_zeros(msg, ombh_len);
|
||
|
||
ombh->type = htons(mb->type);
|
||
ombh->len = htons(ombh_len);
|
||
ombh->rate = htonl(mb->rate);
|
||
ombh->burst_size = htonl(mb->burst_size);
|
||
ombh->prec_level = mb->prec_level;
|
||
|
||
mb++;
|
||
}
|
||
}
|
||
|
||
/* Encode a meter stat for 'mc' and append it to 'replies'. */
|
||
void
|
||
ofputil_append_meter_config(struct list *replies,
|
||
const struct ofputil_meter_config *mc)
|
||
{
|
||
struct ofpbuf *msg = ofpbuf_from_list(list_back(replies));
|
||
size_t start_ofs = ofpbuf_size(msg);
|
||
struct ofp13_meter_config *reply = ofpbuf_put_uninit(msg, sizeof *reply);
|
||
reply->flags = htons(mc->flags);
|
||
reply->meter_id = htonl(mc->meter_id);
|
||
|
||
ofputil_put_bands(mc->n_bands, mc->bands, msg);
|
||
|
||
reply->length = htons(ofpbuf_size(msg) - start_ofs);
|
||
|
||
ofpmp_postappend(replies, start_ofs);
|
||
}
|
||
|
||
/* Encode a meter stat for 'ms' and append it to 'replies'. */
|
||
void
|
||
ofputil_append_meter_stats(struct list *replies,
|
||
const struct ofputil_meter_stats *ms)
|
||
{
|
||
struct ofp13_meter_stats *reply;
|
||
uint16_t n = 0;
|
||
uint16_t len;
|
||
|
||
len = sizeof *reply + ms->n_bands * sizeof(struct ofp13_meter_band_stats);
|
||
reply = ofpmp_append(replies, len);
|
||
|
||
reply->meter_id = htonl(ms->meter_id);
|
||
reply->len = htons(len);
|
||
memset(reply->pad, 0, sizeof reply->pad);
|
||
reply->flow_count = htonl(ms->flow_count);
|
||
reply->packet_in_count = htonll(ms->packet_in_count);
|
||
reply->byte_in_count = htonll(ms->byte_in_count);
|
||
reply->duration_sec = htonl(ms->duration_sec);
|
||
reply->duration_nsec = htonl(ms->duration_nsec);
|
||
|
||
for (n = 0; n < ms->n_bands; ++n) {
|
||
const struct ofputil_meter_band_stats *src = &ms->bands[n];
|
||
struct ofp13_meter_band_stats *dst = &reply->band_stats[n];
|
||
|
||
dst->packet_band_count = htonll(src->packet_count);
|
||
dst->byte_band_count = htonll(src->byte_count);
|
||
}
|
||
}
|
||
|
||
/* Converts an OFPMP_METER_CONFIG reply in 'msg' into an abstract
|
||
* ofputil_meter_config in 'mc', with mc->bands pointing to bands decoded into
|
||
* 'bands'. The caller must have initialized 'bands' and retains ownership of
|
||
* it across the call.
|
||
*
|
||
* Multiple OFPST13_METER_CONFIG replies can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the replies. 'bands' is cleared for each reply.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_meter_config(struct ofpbuf *msg,
|
||
struct ofputil_meter_config *mc,
|
||
struct ofpbuf *bands)
|
||
{
|
||
const struct ofp13_meter_config *omc;
|
||
enum ofperr err;
|
||
|
||
/* Pull OpenFlow headers for the first call. */
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
omc = ofpbuf_try_pull(msg, sizeof *omc);
|
||
if (!omc) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl,
|
||
"OFPMP_METER_CONFIG reply has %"PRIu32" leftover bytes at end",
|
||
ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
ofpbuf_clear(bands);
|
||
err = ofputil_pull_bands(msg, ntohs(omc->length) - sizeof *omc,
|
||
&mc->n_bands, bands);
|
||
if (err) {
|
||
return err;
|
||
}
|
||
mc->meter_id = ntohl(omc->meter_id);
|
||
mc->flags = ntohs(omc->flags);
|
||
mc->bands = ofpbuf_data(bands);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_band_stats(struct ofpbuf *msg, size_t len, uint16_t *n_bands,
|
||
struct ofpbuf *bands)
|
||
{
|
||
const struct ofp13_meter_band_stats *ombs;
|
||
struct ofputil_meter_band_stats *mbs;
|
||
uint16_t n, i;
|
||
|
||
ombs = ofpbuf_try_pull(msg, len);
|
||
if (!ombs) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
n = len / sizeof *ombs;
|
||
if (len != n * sizeof *ombs) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
mbs = ofpbuf_put_uninit(bands, len);
|
||
|
||
for (i = 0; i < n; ++i) {
|
||
mbs[i].packet_count = ntohll(ombs[i].packet_band_count);
|
||
mbs[i].byte_count = ntohll(ombs[i].byte_band_count);
|
||
}
|
||
*n_bands = n;
|
||
return 0;
|
||
}
|
||
|
||
/* Converts an OFPMP_METER reply in 'msg' into an abstract
|
||
* ofputil_meter_stats in 'ms', with ms->bands pointing to band stats
|
||
* decoded into 'bands'.
|
||
*
|
||
* Multiple OFPMP_METER replies can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the replies. 'bands' is cleared for each reply.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_meter_stats(struct ofpbuf *msg,
|
||
struct ofputil_meter_stats *ms,
|
||
struct ofpbuf *bands)
|
||
{
|
||
const struct ofp13_meter_stats *oms;
|
||
enum ofperr err;
|
||
|
||
/* Pull OpenFlow headers for the first call. */
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
oms = ofpbuf_try_pull(msg, sizeof *oms);
|
||
if (!oms) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl,
|
||
"OFPMP_METER reply has %"PRIu32" leftover bytes at end",
|
||
ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
ofpbuf_clear(bands);
|
||
err = ofputil_pull_band_stats(msg, ntohs(oms->len) - sizeof *oms,
|
||
&ms->n_bands, bands);
|
||
if (err) {
|
||
return err;
|
||
}
|
||
ms->meter_id = ntohl(oms->meter_id);
|
||
ms->flow_count = ntohl(oms->flow_count);
|
||
ms->packet_in_count = ntohll(oms->packet_in_count);
|
||
ms->byte_in_count = ntohll(oms->byte_in_count);
|
||
ms->duration_sec = ntohl(oms->duration_sec);
|
||
ms->duration_nsec = ntohl(oms->duration_nsec);
|
||
ms->bands = ofpbuf_data(bands);
|
||
|
||
return 0;
|
||
}
|
||
|
||
void
|
||
ofputil_decode_meter_features(const struct ofp_header *oh,
|
||
struct ofputil_meter_features *mf)
|
||
{
|
||
const struct ofp13_meter_features *omf = ofpmsg_body(oh);
|
||
|
||
mf->max_meters = ntohl(omf->max_meter);
|
||
mf->band_types = ntohl(omf->band_types);
|
||
mf->capabilities = ntohl(omf->capabilities);
|
||
mf->max_bands = omf->max_bands;
|
||
mf->max_color = omf->max_color;
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_meter_features_reply(const struct ofputil_meter_features *mf,
|
||
const struct ofp_header *request)
|
||
{
|
||
struct ofpbuf *reply;
|
||
struct ofp13_meter_features *omf;
|
||
|
||
reply = ofpraw_alloc_stats_reply(request, 0);
|
||
omf = ofpbuf_put_zeros(reply, sizeof *omf);
|
||
|
||
omf->max_meter = htonl(mf->max_meters);
|
||
omf->band_types = htonl(mf->band_types);
|
||
omf->capabilities = htonl(mf->capabilities);
|
||
omf->max_bands = mf->max_bands;
|
||
omf->max_color = mf->max_color;
|
||
|
||
return reply;
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_meter_mod(enum ofp_version ofp_version,
|
||
const struct ofputil_meter_mod *mm)
|
||
{
|
||
struct ofpbuf *msg;
|
||
|
||
struct ofp13_meter_mod *omm;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_OFPT13_METER_MOD, ofp_version,
|
||
NXM_TYPICAL_LEN + mm->meter.n_bands * 16);
|
||
omm = ofpbuf_put_zeros(msg, sizeof *omm);
|
||
omm->command = htons(mm->command);
|
||
if (mm->command != OFPMC13_DELETE) {
|
||
omm->flags = htons(mm->meter.flags);
|
||
}
|
||
omm->meter_id = htonl(mm->meter.meter_id);
|
||
|
||
ofputil_put_bands(mm->meter.n_bands, mm->meter.bands, msg);
|
||
|
||
ofpmsg_update_length(msg);
|
||
return msg;
|
||
}
|
||
|
||
static ovs_be16
|
||
ofputil_tid_command(const struct ofputil_flow_mod *fm,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
return htons(protocol & OFPUTIL_P_TID
|
||
? (fm->command & 0xff) | (fm->table_id << 8)
|
||
: fm->command);
|
||
}
|
||
|
||
/* Converts 'fm' into an OFPT_FLOW_MOD or NXT_FLOW_MOD message according to
|
||
* 'protocol' and returns the message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_flow_mod(const struct ofputil_flow_mod *fm,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
enum ofp_version version = ofputil_protocol_to_ofp_version(protocol);
|
||
ovs_be16 raw_flags = ofputil_encode_flow_mod_flags(fm->flags, version);
|
||
struct ofpbuf *msg;
|
||
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF11_STD:
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM: {
|
||
struct ofp11_flow_mod *ofm;
|
||
int tailroom;
|
||
|
||
tailroom = ofputil_match_typical_len(protocol) + fm->ofpacts_len;
|
||
msg = ofpraw_alloc(OFPRAW_OFPT11_FLOW_MOD, version, tailroom);
|
||
ofm = ofpbuf_put_zeros(msg, sizeof *ofm);
|
||
if ((protocol == OFPUTIL_P_OF11_STD
|
||
&& (fm->command == OFPFC_MODIFY ||
|
||
fm->command == OFPFC_MODIFY_STRICT)
|
||
&& fm->cookie_mask == htonll(0))
|
||
|| fm->command == OFPFC_ADD) {
|
||
ofm->cookie = fm->new_cookie;
|
||
} else {
|
||
ofm->cookie = fm->cookie;
|
||
}
|
||
ofm->cookie_mask = fm->cookie_mask;
|
||
if (fm->table_id != OFPTT_ALL
|
||
|| (protocol != OFPUTIL_P_OF11_STD
|
||
&& (fm->command == OFPFC_DELETE ||
|
||
fm->command == OFPFC_DELETE_STRICT))) {
|
||
ofm->table_id = fm->table_id;
|
||
} else {
|
||
ofm->table_id = 0;
|
||
}
|
||
ofm->command = fm->command;
|
||
ofm->idle_timeout = htons(fm->idle_timeout);
|
||
ofm->hard_timeout = htons(fm->hard_timeout);
|
||
ofm->priority = htons(fm->priority);
|
||
ofm->buffer_id = htonl(fm->buffer_id);
|
||
ofm->out_port = ofputil_port_to_ofp11(fm->out_port);
|
||
ofm->out_group = htonl(fm->out_group);
|
||
ofm->flags = raw_flags;
|
||
ofputil_put_ofp11_match(msg, &fm->match, protocol);
|
||
ofpacts_put_openflow_instructions(fm->ofpacts, fm->ofpacts_len, msg,
|
||
version);
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID: {
|
||
struct ofp10_flow_mod *ofm;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_OFPT10_FLOW_MOD, OFP10_VERSION,
|
||
fm->ofpacts_len);
|
||
ofm = ofpbuf_put_zeros(msg, sizeof *ofm);
|
||
ofputil_match_to_ofp10_match(&fm->match, &ofm->match);
|
||
ofm->cookie = fm->new_cookie;
|
||
ofm->command = ofputil_tid_command(fm, protocol);
|
||
ofm->idle_timeout = htons(fm->idle_timeout);
|
||
ofm->hard_timeout = htons(fm->hard_timeout);
|
||
ofm->priority = htons(fm->priority);
|
||
ofm->buffer_id = htonl(fm->buffer_id);
|
||
ofm->out_port = htons(ofp_to_u16(fm->out_port));
|
||
ofm->flags = raw_flags;
|
||
ofpacts_put_openflow_actions(fm->ofpacts, fm->ofpacts_len, msg,
|
||
version);
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID: {
|
||
struct nx_flow_mod *nfm;
|
||
int match_len;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_NXT_FLOW_MOD, OFP10_VERSION,
|
||
NXM_TYPICAL_LEN + fm->ofpacts_len);
|
||
nfm = ofpbuf_put_zeros(msg, sizeof *nfm);
|
||
nfm->command = ofputil_tid_command(fm, protocol);
|
||
nfm->cookie = fm->new_cookie;
|
||
match_len = nx_put_match(msg, &fm->match, fm->cookie, fm->cookie_mask);
|
||
nfm = ofpbuf_l3(msg);
|
||
nfm->idle_timeout = htons(fm->idle_timeout);
|
||
nfm->hard_timeout = htons(fm->hard_timeout);
|
||
nfm->priority = htons(fm->priority);
|
||
nfm->buffer_id = htonl(fm->buffer_id);
|
||
nfm->out_port = htons(ofp_to_u16(fm->out_port));
|
||
nfm->flags = raw_flags;
|
||
nfm->match_len = htons(match_len);
|
||
ofpacts_put_openflow_actions(fm->ofpacts, fm->ofpacts_len, msg,
|
||
version);
|
||
break;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
ofpmsg_update_length(msg);
|
||
return msg;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_decode_ofpst10_flow_request(struct ofputil_flow_stats_request *fsr,
|
||
const struct ofp10_flow_stats_request *ofsr,
|
||
bool aggregate)
|
||
{
|
||
fsr->aggregate = aggregate;
|
||
ofputil_match_from_ofp10_match(&ofsr->match, &fsr->match);
|
||
fsr->out_port = u16_to_ofp(ntohs(ofsr->out_port));
|
||
fsr->out_group = OFPG11_ANY;
|
||
fsr->table_id = ofsr->table_id;
|
||
fsr->cookie = fsr->cookie_mask = htonll(0);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_decode_ofpst11_flow_request(struct ofputil_flow_stats_request *fsr,
|
||
struct ofpbuf *b, bool aggregate)
|
||
{
|
||
const struct ofp11_flow_stats_request *ofsr;
|
||
enum ofperr error;
|
||
|
||
ofsr = ofpbuf_pull(b, sizeof *ofsr);
|
||
fsr->aggregate = aggregate;
|
||
fsr->table_id = ofsr->table_id;
|
||
error = ofputil_port_from_ofp11(ofsr->out_port, &fsr->out_port);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
fsr->out_group = ntohl(ofsr->out_group);
|
||
fsr->cookie = ofsr->cookie;
|
||
fsr->cookie_mask = ofsr->cookie_mask;
|
||
error = ofputil_pull_ofp11_match(b, &fsr->match, NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_decode_nxst_flow_request(struct ofputil_flow_stats_request *fsr,
|
||
struct ofpbuf *b, bool aggregate)
|
||
{
|
||
const struct nx_flow_stats_request *nfsr;
|
||
enum ofperr error;
|
||
|
||
nfsr = ofpbuf_pull(b, sizeof *nfsr);
|
||
error = nx_pull_match(b, ntohs(nfsr->match_len), &fsr->match,
|
||
&fsr->cookie, &fsr->cookie_mask);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
if (ofpbuf_size(b)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
fsr->aggregate = aggregate;
|
||
fsr->out_port = u16_to_ofp(ntohs(nfsr->out_port));
|
||
fsr->out_group = OFPG11_ANY;
|
||
fsr->table_id = nfsr->table_id;
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Constructs and returns an OFPT_QUEUE_GET_CONFIG request for the specified
|
||
* 'port', suitable for OpenFlow version 'version'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_queue_get_config_request(enum ofp_version version,
|
||
ofp_port_t port)
|
||
{
|
||
struct ofpbuf *request;
|
||
|
||
if (version == OFP10_VERSION) {
|
||
struct ofp10_queue_get_config_request *qgcr10;
|
||
|
||
request = ofpraw_alloc(OFPRAW_OFPT10_QUEUE_GET_CONFIG_REQUEST,
|
||
version, 0);
|
||
qgcr10 = ofpbuf_put_zeros(request, sizeof *qgcr10);
|
||
qgcr10->port = htons(ofp_to_u16(port));
|
||
} else {
|
||
struct ofp11_queue_get_config_request *qgcr11;
|
||
|
||
request = ofpraw_alloc(OFPRAW_OFPT11_QUEUE_GET_CONFIG_REQUEST,
|
||
version, 0);
|
||
qgcr11 = ofpbuf_put_zeros(request, sizeof *qgcr11);
|
||
qgcr11->port = ofputil_port_to_ofp11(port);
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
/* Parses OFPT_QUEUE_GET_CONFIG request 'oh', storing the port specified by the
|
||
* request into '*port'. Returns 0 if successful, otherwise an OpenFlow error
|
||
* code. */
|
||
enum ofperr
|
||
ofputil_decode_queue_get_config_request(const struct ofp_header *oh,
|
||
ofp_port_t *port)
|
||
{
|
||
const struct ofp10_queue_get_config_request *qgcr10;
|
||
const struct ofp11_queue_get_config_request *qgcr11;
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
switch ((int) raw) {
|
||
case OFPRAW_OFPT10_QUEUE_GET_CONFIG_REQUEST:
|
||
qgcr10 = ofpbuf_data(&b);
|
||
*port = u16_to_ofp(ntohs(qgcr10->port));
|
||
return 0;
|
||
|
||
case OFPRAW_OFPT11_QUEUE_GET_CONFIG_REQUEST:
|
||
qgcr11 = ofpbuf_data(&b);
|
||
return ofputil_port_from_ofp11(qgcr11->port, port);
|
||
}
|
||
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
/* Constructs and returns the beginning of a reply to
|
||
* OFPT_QUEUE_GET_CONFIG_REQUEST 'oh'. The caller may append information about
|
||
* individual queues with ofputil_append_queue_get_config_reply(). */
|
||
struct ofpbuf *
|
||
ofputil_encode_queue_get_config_reply(const struct ofp_header *oh)
|
||
{
|
||
struct ofp10_queue_get_config_reply *qgcr10;
|
||
struct ofp11_queue_get_config_reply *qgcr11;
|
||
struct ofpbuf *reply;
|
||
enum ofperr error;
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
ofp_port_t port;
|
||
|
||
error = ofputil_decode_queue_get_config_request(oh, &port);
|
||
ovs_assert(!error);
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
switch ((int) raw) {
|
||
case OFPRAW_OFPT10_QUEUE_GET_CONFIG_REQUEST:
|
||
reply = ofpraw_alloc_reply(OFPRAW_OFPT10_QUEUE_GET_CONFIG_REPLY,
|
||
oh, 0);
|
||
qgcr10 = ofpbuf_put_zeros(reply, sizeof *qgcr10);
|
||
qgcr10->port = htons(ofp_to_u16(port));
|
||
break;
|
||
|
||
case OFPRAW_OFPT11_QUEUE_GET_CONFIG_REQUEST:
|
||
reply = ofpraw_alloc_reply(OFPRAW_OFPT11_QUEUE_GET_CONFIG_REPLY,
|
||
oh, 0);
|
||
qgcr11 = ofpbuf_put_zeros(reply, sizeof *qgcr11);
|
||
qgcr11->port = ofputil_port_to_ofp11(port);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return reply;
|
||
}
|
||
|
||
static void
|
||
put_queue_rate(struct ofpbuf *reply, enum ofp_queue_properties property,
|
||
uint16_t rate)
|
||
{
|
||
if (rate != UINT16_MAX) {
|
||
struct ofp_queue_prop_rate *oqpr;
|
||
|
||
oqpr = ofpbuf_put_zeros(reply, sizeof *oqpr);
|
||
oqpr->prop_header.property = htons(property);
|
||
oqpr->prop_header.len = htons(sizeof *oqpr);
|
||
oqpr->rate = htons(rate);
|
||
}
|
||
}
|
||
|
||
/* Appends a queue description for 'queue_id' to the
|
||
* OFPT_QUEUE_GET_CONFIG_REPLY already in 'oh'. */
|
||
void
|
||
ofputil_append_queue_get_config_reply(struct ofpbuf *reply,
|
||
const struct ofputil_queue_config *oqc)
|
||
{
|
||
const struct ofp_header *oh = ofpbuf_data(reply);
|
||
size_t start_ofs, len_ofs;
|
||
ovs_be16 *len;
|
||
|
||
start_ofs = ofpbuf_size(reply);
|
||
if (oh->version < OFP12_VERSION) {
|
||
struct ofp10_packet_queue *opq10;
|
||
|
||
opq10 = ofpbuf_put_zeros(reply, sizeof *opq10);
|
||
opq10->queue_id = htonl(oqc->queue_id);
|
||
len_ofs = (char *) &opq10->len - (char *) ofpbuf_data(reply);
|
||
} else {
|
||
struct ofp11_queue_get_config_reply *qgcr11;
|
||
struct ofp12_packet_queue *opq12;
|
||
ovs_be32 port;
|
||
|
||
qgcr11 = ofpbuf_l3(reply);
|
||
port = qgcr11->port;
|
||
|
||
opq12 = ofpbuf_put_zeros(reply, sizeof *opq12);
|
||
opq12->port = port;
|
||
opq12->queue_id = htonl(oqc->queue_id);
|
||
len_ofs = (char *) &opq12->len - (char *) ofpbuf_data(reply);
|
||
}
|
||
|
||
put_queue_rate(reply, OFPQT_MIN_RATE, oqc->min_rate);
|
||
put_queue_rate(reply, OFPQT_MAX_RATE, oqc->max_rate);
|
||
|
||
len = ofpbuf_at(reply, len_ofs, sizeof *len);
|
||
*len = htons(ofpbuf_size(reply) - start_ofs);
|
||
}
|
||
|
||
/* Decodes the initial part of an OFPT_QUEUE_GET_CONFIG_REPLY from 'reply' and
|
||
* stores in '*port' the port that the reply is about. The caller may call
|
||
* ofputil_pull_queue_get_config_reply() to obtain information about individual
|
||
* queues included in the reply. Returns 0 if successful, otherwise an
|
||
* ofperr.*/
|
||
enum ofperr
|
||
ofputil_decode_queue_get_config_reply(struct ofpbuf *reply, ofp_port_t *port)
|
||
{
|
||
const struct ofp10_queue_get_config_reply *qgcr10;
|
||
const struct ofp11_queue_get_config_reply *qgcr11;
|
||
enum ofpraw raw;
|
||
|
||
raw = ofpraw_pull_assert(reply);
|
||
switch ((int) raw) {
|
||
case OFPRAW_OFPT10_QUEUE_GET_CONFIG_REPLY:
|
||
qgcr10 = ofpbuf_pull(reply, sizeof *qgcr10);
|
||
*port = u16_to_ofp(ntohs(qgcr10->port));
|
||
return 0;
|
||
|
||
case OFPRAW_OFPT11_QUEUE_GET_CONFIG_REPLY:
|
||
qgcr11 = ofpbuf_pull(reply, sizeof *qgcr11);
|
||
return ofputil_port_from_ofp11(qgcr11->port, port);
|
||
}
|
||
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_queue_rate(const struct ofp_queue_prop_header *hdr, uint16_t *rate)
|
||
{
|
||
const struct ofp_queue_prop_rate *oqpr;
|
||
|
||
if (hdr->len == htons(sizeof *oqpr)) {
|
||
oqpr = (const struct ofp_queue_prop_rate *) hdr;
|
||
*rate = ntohs(oqpr->rate);
|
||
return 0;
|
||
} else {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
}
|
||
|
||
/* Decodes information about a queue from the OFPT_QUEUE_GET_CONFIG_REPLY in
|
||
* 'reply' and stores it in '*queue'. ofputil_decode_queue_get_config_reply()
|
||
* must already have pulled off the main header.
|
||
*
|
||
* This function returns EOF if the last queue has already been decoded, 0 if a
|
||
* queue was successfully decoded into '*queue', or an ofperr if there was a
|
||
* problem decoding 'reply'. */
|
||
int
|
||
ofputil_pull_queue_get_config_reply(struct ofpbuf *reply,
|
||
struct ofputil_queue_config *queue)
|
||
{
|
||
const struct ofp_header *oh;
|
||
unsigned int opq_len;
|
||
unsigned int len;
|
||
|
||
if (!ofpbuf_size(reply)) {
|
||
return EOF;
|
||
}
|
||
|
||
queue->min_rate = UINT16_MAX;
|
||
queue->max_rate = UINT16_MAX;
|
||
|
||
oh = reply->frame;
|
||
if (oh->version < OFP12_VERSION) {
|
||
const struct ofp10_packet_queue *opq10;
|
||
|
||
opq10 = ofpbuf_try_pull(reply, sizeof *opq10);
|
||
if (!opq10) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
queue->queue_id = ntohl(opq10->queue_id);
|
||
len = ntohs(opq10->len);
|
||
opq_len = sizeof *opq10;
|
||
} else {
|
||
const struct ofp12_packet_queue *opq12;
|
||
|
||
opq12 = ofpbuf_try_pull(reply, sizeof *opq12);
|
||
if (!opq12) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
queue->queue_id = ntohl(opq12->queue_id);
|
||
len = ntohs(opq12->len);
|
||
opq_len = sizeof *opq12;
|
||
}
|
||
|
||
if (len < opq_len || len > ofpbuf_size(reply) + opq_len || len % 8) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
len -= opq_len;
|
||
|
||
while (len > 0) {
|
||
const struct ofp_queue_prop_header *hdr;
|
||
unsigned int property;
|
||
unsigned int prop_len;
|
||
enum ofperr error = 0;
|
||
|
||
hdr = ofpbuf_at_assert(reply, 0, sizeof *hdr);
|
||
prop_len = ntohs(hdr->len);
|
||
if (prop_len < sizeof *hdr || prop_len > ofpbuf_size(reply) || prop_len % 8) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
property = ntohs(hdr->property);
|
||
switch (property) {
|
||
case OFPQT_MIN_RATE:
|
||
error = parse_queue_rate(hdr, &queue->min_rate);
|
||
break;
|
||
|
||
case OFPQT_MAX_RATE:
|
||
error = parse_queue_rate(hdr, &queue->max_rate);
|
||
break;
|
||
|
||
default:
|
||
VLOG_INFO_RL(&bad_ofmsg_rl, "unknown queue property %u", property);
|
||
break;
|
||
}
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
ofpbuf_pull(reply, prop_len);
|
||
len -= prop_len;
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
/* Converts an OFPST_FLOW, OFPST_AGGREGATE, NXST_FLOW, or NXST_AGGREGATE
|
||
* request 'oh', into an abstract flow_stats_request in 'fsr'. Returns 0 if
|
||
* successful, otherwise an OpenFlow error code. */
|
||
enum ofperr
|
||
ofputil_decode_flow_stats_request(struct ofputil_flow_stats_request *fsr,
|
||
const struct ofp_header *oh)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
switch ((int) raw) {
|
||
case OFPRAW_OFPST10_FLOW_REQUEST:
|
||
return ofputil_decode_ofpst10_flow_request(fsr, ofpbuf_data(&b), false);
|
||
|
||
case OFPRAW_OFPST10_AGGREGATE_REQUEST:
|
||
return ofputil_decode_ofpst10_flow_request(fsr, ofpbuf_data(&b), true);
|
||
|
||
case OFPRAW_OFPST11_FLOW_REQUEST:
|
||
return ofputil_decode_ofpst11_flow_request(fsr, &b, false);
|
||
|
||
case OFPRAW_OFPST11_AGGREGATE_REQUEST:
|
||
return ofputil_decode_ofpst11_flow_request(fsr, &b, true);
|
||
|
||
case OFPRAW_NXST_FLOW_REQUEST:
|
||
return ofputil_decode_nxst_flow_request(fsr, &b, false);
|
||
|
||
case OFPRAW_NXST_AGGREGATE_REQUEST:
|
||
return ofputil_decode_nxst_flow_request(fsr, &b, true);
|
||
|
||
default:
|
||
/* Hey, the caller lied. */
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Converts abstract flow_stats_request 'fsr' into an OFPST_FLOW,
|
||
* OFPST_AGGREGATE, NXST_FLOW, or NXST_AGGREGATE request 'oh' according to
|
||
* 'protocol', and returns the message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_flow_stats_request(const struct ofputil_flow_stats_request *fsr,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
struct ofpbuf *msg;
|
||
enum ofpraw raw;
|
||
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF11_STD:
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM: {
|
||
struct ofp11_flow_stats_request *ofsr;
|
||
|
||
raw = (fsr->aggregate
|
||
? OFPRAW_OFPST11_AGGREGATE_REQUEST
|
||
: OFPRAW_OFPST11_FLOW_REQUEST);
|
||
msg = ofpraw_alloc(raw, ofputil_protocol_to_ofp_version(protocol),
|
||
ofputil_match_typical_len(protocol));
|
||
ofsr = ofpbuf_put_zeros(msg, sizeof *ofsr);
|
||
ofsr->table_id = fsr->table_id;
|
||
ofsr->out_port = ofputil_port_to_ofp11(fsr->out_port);
|
||
ofsr->out_group = htonl(fsr->out_group);
|
||
ofsr->cookie = fsr->cookie;
|
||
ofsr->cookie_mask = fsr->cookie_mask;
|
||
ofputil_put_ofp11_match(msg, &fsr->match, protocol);
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID: {
|
||
struct ofp10_flow_stats_request *ofsr;
|
||
|
||
raw = (fsr->aggregate
|
||
? OFPRAW_OFPST10_AGGREGATE_REQUEST
|
||
: OFPRAW_OFPST10_FLOW_REQUEST);
|
||
msg = ofpraw_alloc(raw, OFP10_VERSION, 0);
|
||
ofsr = ofpbuf_put_zeros(msg, sizeof *ofsr);
|
||
ofputil_match_to_ofp10_match(&fsr->match, &ofsr->match);
|
||
ofsr->table_id = fsr->table_id;
|
||
ofsr->out_port = htons(ofp_to_u16(fsr->out_port));
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID: {
|
||
struct nx_flow_stats_request *nfsr;
|
||
int match_len;
|
||
|
||
raw = (fsr->aggregate
|
||
? OFPRAW_NXST_AGGREGATE_REQUEST
|
||
: OFPRAW_NXST_FLOW_REQUEST);
|
||
msg = ofpraw_alloc(raw, OFP10_VERSION, NXM_TYPICAL_LEN);
|
||
ofpbuf_put_zeros(msg, sizeof *nfsr);
|
||
match_len = nx_put_match(msg, &fsr->match,
|
||
fsr->cookie, fsr->cookie_mask);
|
||
|
||
nfsr = ofpbuf_l3(msg);
|
||
nfsr->out_port = htons(ofp_to_u16(fsr->out_port));
|
||
nfsr->match_len = htons(match_len);
|
||
nfsr->table_id = fsr->table_id;
|
||
break;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return msg;
|
||
}
|
||
|
||
/* Converts an OFPST_FLOW or NXST_FLOW reply in 'msg' into an abstract
|
||
* ofputil_flow_stats in 'fs'.
|
||
*
|
||
* Multiple OFPST_FLOW or NXST_FLOW replies can be packed into a single
|
||
* OpenFlow message. Calling this function multiple times for a single 'msg'
|
||
* iterates through the replies. The caller must initially leave 'msg''s layer
|
||
* pointers null and not modify them between calls.
|
||
*
|
||
* Most switches don't send the values needed to populate fs->idle_age and
|
||
* fs->hard_age, so those members will usually be set to 0. If the switch from
|
||
* which 'msg' originated is known to implement NXT_FLOW_AGE, then pass
|
||
* 'flow_age_extension' as true so that the contents of 'msg' determine the
|
||
* 'idle_age' and 'hard_age' members in 'fs'.
|
||
*
|
||
* Uses 'ofpacts' to store the abstract OFPACT_* version of the flow stats
|
||
* reply's actions. The caller must initialize 'ofpacts' and retains ownership
|
||
* of it. 'fs->ofpacts' will point into the 'ofpacts' buffer.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_flow_stats_reply(struct ofputil_flow_stats *fs,
|
||
struct ofpbuf *msg,
|
||
bool flow_age_extension,
|
||
struct ofpbuf *ofpacts)
|
||
{
|
||
const struct ofp_header *oh;
|
||
size_t instructions_len;
|
||
enum ofperr error;
|
||
enum ofpraw raw;
|
||
|
||
error = (msg->frame
|
||
? ofpraw_decode(&raw, msg->frame)
|
||
: ofpraw_pull(&raw, msg));
|
||
if (error) {
|
||
return error;
|
||
}
|
||
oh = msg->frame;
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
} else if (raw == OFPRAW_OFPST11_FLOW_REPLY
|
||
|| raw == OFPRAW_OFPST13_FLOW_REPLY) {
|
||
const struct ofp11_flow_stats *ofs;
|
||
size_t length;
|
||
uint16_t padded_match_len;
|
||
|
||
ofs = ofpbuf_try_pull(msg, sizeof *ofs);
|
||
if (!ofs) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply has %"PRIu32" leftover "
|
||
"bytes at end", ofpbuf_size(msg));
|
||
return EINVAL;
|
||
}
|
||
|
||
length = ntohs(ofs->length);
|
||
if (length < sizeof *ofs) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply claims invalid "
|
||
"length %"PRIuSIZE, length);
|
||
return EINVAL;
|
||
}
|
||
|
||
if (ofputil_pull_ofp11_match(msg, &fs->match, &padded_match_len)) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply bad match");
|
||
return EINVAL;
|
||
}
|
||
instructions_len = length - sizeof *ofs - padded_match_len;
|
||
|
||
fs->priority = ntohs(ofs->priority);
|
||
fs->table_id = ofs->table_id;
|
||
fs->duration_sec = ntohl(ofs->duration_sec);
|
||
fs->duration_nsec = ntohl(ofs->duration_nsec);
|
||
fs->idle_timeout = ntohs(ofs->idle_timeout);
|
||
fs->hard_timeout = ntohs(ofs->hard_timeout);
|
||
if (raw == OFPRAW_OFPST13_FLOW_REPLY) {
|
||
error = ofputil_decode_flow_mod_flags(ofs->flags, -1, oh->version,
|
||
&fs->flags);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
} else {
|
||
fs->flags = 0;
|
||
}
|
||
fs->idle_age = -1;
|
||
fs->hard_age = -1;
|
||
fs->cookie = ofs->cookie;
|
||
fs->packet_count = ntohll(ofs->packet_count);
|
||
fs->byte_count = ntohll(ofs->byte_count);
|
||
} else if (raw == OFPRAW_OFPST10_FLOW_REPLY) {
|
||
const struct ofp10_flow_stats *ofs;
|
||
size_t length;
|
||
|
||
ofs = ofpbuf_try_pull(msg, sizeof *ofs);
|
||
if (!ofs) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply has %"PRIu32" leftover "
|
||
"bytes at end", ofpbuf_size(msg));
|
||
return EINVAL;
|
||
}
|
||
|
||
length = ntohs(ofs->length);
|
||
if (length < sizeof *ofs) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply claims invalid "
|
||
"length %"PRIuSIZE, length);
|
||
return EINVAL;
|
||
}
|
||
instructions_len = length - sizeof *ofs;
|
||
|
||
fs->cookie = get_32aligned_be64(&ofs->cookie);
|
||
ofputil_match_from_ofp10_match(&ofs->match, &fs->match);
|
||
fs->priority = ntohs(ofs->priority);
|
||
fs->table_id = ofs->table_id;
|
||
fs->duration_sec = ntohl(ofs->duration_sec);
|
||
fs->duration_nsec = ntohl(ofs->duration_nsec);
|
||
fs->idle_timeout = ntohs(ofs->idle_timeout);
|
||
fs->hard_timeout = ntohs(ofs->hard_timeout);
|
||
fs->idle_age = -1;
|
||
fs->hard_age = -1;
|
||
fs->packet_count = ntohll(get_32aligned_be64(&ofs->packet_count));
|
||
fs->byte_count = ntohll(get_32aligned_be64(&ofs->byte_count));
|
||
fs->flags = 0;
|
||
} else if (raw == OFPRAW_NXST_FLOW_REPLY) {
|
||
const struct nx_flow_stats *nfs;
|
||
size_t match_len, length;
|
||
|
||
nfs = ofpbuf_try_pull(msg, sizeof *nfs);
|
||
if (!nfs) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "NXST_FLOW reply has %"PRIu32" leftover "
|
||
"bytes at end", ofpbuf_size(msg));
|
||
return EINVAL;
|
||
}
|
||
|
||
length = ntohs(nfs->length);
|
||
match_len = ntohs(nfs->match_len);
|
||
if (length < sizeof *nfs + ROUND_UP(match_len, 8)) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "NXST_FLOW reply with match_len=%"PRIuSIZE" "
|
||
"claims invalid length %"PRIuSIZE, match_len, length);
|
||
return EINVAL;
|
||
}
|
||
if (nx_pull_match(msg, match_len, &fs->match, NULL, NULL)) {
|
||
return EINVAL;
|
||
}
|
||
instructions_len = length - sizeof *nfs - ROUND_UP(match_len, 8);
|
||
|
||
fs->cookie = nfs->cookie;
|
||
fs->table_id = nfs->table_id;
|
||
fs->duration_sec = ntohl(nfs->duration_sec);
|
||
fs->duration_nsec = ntohl(nfs->duration_nsec);
|
||
fs->priority = ntohs(nfs->priority);
|
||
fs->idle_timeout = ntohs(nfs->idle_timeout);
|
||
fs->hard_timeout = ntohs(nfs->hard_timeout);
|
||
fs->idle_age = -1;
|
||
fs->hard_age = -1;
|
||
if (flow_age_extension) {
|
||
if (nfs->idle_age) {
|
||
fs->idle_age = ntohs(nfs->idle_age) - 1;
|
||
}
|
||
if (nfs->hard_age) {
|
||
fs->hard_age = ntohs(nfs->hard_age) - 1;
|
||
}
|
||
}
|
||
fs->packet_count = ntohll(nfs->packet_count);
|
||
fs->byte_count = ntohll(nfs->byte_count);
|
||
fs->flags = 0;
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
if (ofpacts_pull_openflow_instructions(msg, instructions_len, oh->version,
|
||
ofpacts)) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_FLOW reply bad instructions");
|
||
return EINVAL;
|
||
}
|
||
fs->ofpacts = ofpbuf_data(ofpacts);
|
||
fs->ofpacts_len = ofpbuf_size(ofpacts);
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Returns 'count' unchanged except that UINT64_MAX becomes 0.
|
||
*
|
||
* We use this in situations where OVS internally uses UINT64_MAX to mean
|
||
* "value unknown" but OpenFlow 1.0 does not define any unknown value. */
|
||
static uint64_t
|
||
unknown_to_zero(uint64_t count)
|
||
{
|
||
return count != UINT64_MAX ? count : 0;
|
||
}
|
||
|
||
/* Appends an OFPST_FLOW or NXST_FLOW reply that contains the data in 'fs' to
|
||
* those already present in the list of ofpbufs in 'replies'. 'replies' should
|
||
* have been initialized with ofpmp_init(). */
|
||
void
|
||
ofputil_append_flow_stats_reply(const struct ofputil_flow_stats *fs,
|
||
struct list *replies)
|
||
{
|
||
struct ofpbuf *reply = ofpbuf_from_list(list_back(replies));
|
||
size_t start_ofs = ofpbuf_size(reply);
|
||
enum ofp_version version = ofpmp_version(replies);
|
||
enum ofpraw raw = ofpmp_decode_raw(replies);
|
||
|
||
if (raw == OFPRAW_OFPST11_FLOW_REPLY || raw == OFPRAW_OFPST13_FLOW_REPLY) {
|
||
struct ofp11_flow_stats *ofs;
|
||
|
||
ofpbuf_put_uninit(reply, sizeof *ofs);
|
||
oxm_put_match(reply, &fs->match, version);
|
||
ofpacts_put_openflow_instructions(fs->ofpacts, fs->ofpacts_len, reply,
|
||
version);
|
||
|
||
ofs = ofpbuf_at_assert(reply, start_ofs, sizeof *ofs);
|
||
ofs->length = htons(ofpbuf_size(reply) - start_ofs);
|
||
ofs->table_id = fs->table_id;
|
||
ofs->pad = 0;
|
||
ofs->duration_sec = htonl(fs->duration_sec);
|
||
ofs->duration_nsec = htonl(fs->duration_nsec);
|
||
ofs->priority = htons(fs->priority);
|
||
ofs->idle_timeout = htons(fs->idle_timeout);
|
||
ofs->hard_timeout = htons(fs->hard_timeout);
|
||
if (raw == OFPRAW_OFPST13_FLOW_REPLY) {
|
||
ofs->flags = ofputil_encode_flow_mod_flags(fs->flags, version);
|
||
} else {
|
||
ofs->flags = 0;
|
||
}
|
||
memset(ofs->pad2, 0, sizeof ofs->pad2);
|
||
ofs->cookie = fs->cookie;
|
||
ofs->packet_count = htonll(unknown_to_zero(fs->packet_count));
|
||
ofs->byte_count = htonll(unknown_to_zero(fs->byte_count));
|
||
} else if (raw == OFPRAW_OFPST10_FLOW_REPLY) {
|
||
struct ofp10_flow_stats *ofs;
|
||
|
||
ofpbuf_put_uninit(reply, sizeof *ofs);
|
||
ofpacts_put_openflow_actions(fs->ofpacts, fs->ofpacts_len, reply,
|
||
version);
|
||
ofs = ofpbuf_at_assert(reply, start_ofs, sizeof *ofs);
|
||
ofs->length = htons(ofpbuf_size(reply) - start_ofs);
|
||
ofs->table_id = fs->table_id;
|
||
ofs->pad = 0;
|
||
ofputil_match_to_ofp10_match(&fs->match, &ofs->match);
|
||
ofs->duration_sec = htonl(fs->duration_sec);
|
||
ofs->duration_nsec = htonl(fs->duration_nsec);
|
||
ofs->priority = htons(fs->priority);
|
||
ofs->idle_timeout = htons(fs->idle_timeout);
|
||
ofs->hard_timeout = htons(fs->hard_timeout);
|
||
memset(ofs->pad2, 0, sizeof ofs->pad2);
|
||
put_32aligned_be64(&ofs->cookie, fs->cookie);
|
||
put_32aligned_be64(&ofs->packet_count,
|
||
htonll(unknown_to_zero(fs->packet_count)));
|
||
put_32aligned_be64(&ofs->byte_count,
|
||
htonll(unknown_to_zero(fs->byte_count)));
|
||
} else if (raw == OFPRAW_NXST_FLOW_REPLY) {
|
||
struct nx_flow_stats *nfs;
|
||
int match_len;
|
||
|
||
ofpbuf_put_uninit(reply, sizeof *nfs);
|
||
match_len = nx_put_match(reply, &fs->match, 0, 0);
|
||
ofpacts_put_openflow_actions(fs->ofpacts, fs->ofpacts_len, reply,
|
||
version);
|
||
nfs = ofpbuf_at_assert(reply, start_ofs, sizeof *nfs);
|
||
nfs->length = htons(ofpbuf_size(reply) - start_ofs);
|
||
nfs->table_id = fs->table_id;
|
||
nfs->pad = 0;
|
||
nfs->duration_sec = htonl(fs->duration_sec);
|
||
nfs->duration_nsec = htonl(fs->duration_nsec);
|
||
nfs->priority = htons(fs->priority);
|
||
nfs->idle_timeout = htons(fs->idle_timeout);
|
||
nfs->hard_timeout = htons(fs->hard_timeout);
|
||
nfs->idle_age = htons(fs->idle_age < 0 ? 0
|
||
: fs->idle_age < UINT16_MAX ? fs->idle_age + 1
|
||
: UINT16_MAX);
|
||
nfs->hard_age = htons(fs->hard_age < 0 ? 0
|
||
: fs->hard_age < UINT16_MAX ? fs->hard_age + 1
|
||
: UINT16_MAX);
|
||
nfs->match_len = htons(match_len);
|
||
nfs->cookie = fs->cookie;
|
||
nfs->packet_count = htonll(fs->packet_count);
|
||
nfs->byte_count = htonll(fs->byte_count);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
ofpmp_postappend(replies, start_ofs);
|
||
}
|
||
|
||
/* Converts abstract ofputil_aggregate_stats 'stats' into an OFPST_AGGREGATE or
|
||
* NXST_AGGREGATE reply matching 'request', and returns the message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_aggregate_stats_reply(
|
||
const struct ofputil_aggregate_stats *stats,
|
||
const struct ofp_header *request)
|
||
{
|
||
struct ofp_aggregate_stats_reply *asr;
|
||
uint64_t packet_count;
|
||
uint64_t byte_count;
|
||
struct ofpbuf *msg;
|
||
enum ofpraw raw;
|
||
|
||
ofpraw_decode(&raw, request);
|
||
if (raw == OFPRAW_OFPST10_AGGREGATE_REQUEST) {
|
||
packet_count = unknown_to_zero(stats->packet_count);
|
||
byte_count = unknown_to_zero(stats->byte_count);
|
||
} else {
|
||
packet_count = stats->packet_count;
|
||
byte_count = stats->byte_count;
|
||
}
|
||
|
||
msg = ofpraw_alloc_stats_reply(request, 0);
|
||
asr = ofpbuf_put_zeros(msg, sizeof *asr);
|
||
put_32aligned_be64(&asr->packet_count, htonll(packet_count));
|
||
put_32aligned_be64(&asr->byte_count, htonll(byte_count));
|
||
asr->flow_count = htonl(stats->flow_count);
|
||
|
||
return msg;
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_aggregate_stats_reply(struct ofputil_aggregate_stats *stats,
|
||
const struct ofp_header *reply)
|
||
{
|
||
struct ofp_aggregate_stats_reply *asr;
|
||
struct ofpbuf msg;
|
||
|
||
ofpbuf_use_const(&msg, reply, ntohs(reply->length));
|
||
ofpraw_pull_assert(&msg);
|
||
|
||
asr = ofpbuf_l3(&msg);
|
||
stats->packet_count = ntohll(get_32aligned_be64(&asr->packet_count));
|
||
stats->byte_count = ntohll(get_32aligned_be64(&asr->byte_count));
|
||
stats->flow_count = ntohl(asr->flow_count);
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Converts an OFPT_FLOW_REMOVED or NXT_FLOW_REMOVED message 'oh' into an
|
||
* abstract ofputil_flow_removed in 'fr'. Returns 0 if successful, otherwise
|
||
* an OpenFlow error code. */
|
||
enum ofperr
|
||
ofputil_decode_flow_removed(struct ofputil_flow_removed *fr,
|
||
const struct ofp_header *oh)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
if (raw == OFPRAW_OFPT11_FLOW_REMOVED) {
|
||
const struct ofp12_flow_removed *ofr;
|
||
enum ofperr error;
|
||
|
||
ofr = ofpbuf_pull(&b, sizeof *ofr);
|
||
|
||
error = ofputil_pull_ofp11_match(&b, &fr->match, NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
fr->priority = ntohs(ofr->priority);
|
||
fr->cookie = ofr->cookie;
|
||
fr->reason = ofr->reason;
|
||
fr->table_id = ofr->table_id;
|
||
fr->duration_sec = ntohl(ofr->duration_sec);
|
||
fr->duration_nsec = ntohl(ofr->duration_nsec);
|
||
fr->idle_timeout = ntohs(ofr->idle_timeout);
|
||
fr->hard_timeout = ntohs(ofr->hard_timeout);
|
||
fr->packet_count = ntohll(ofr->packet_count);
|
||
fr->byte_count = ntohll(ofr->byte_count);
|
||
} else if (raw == OFPRAW_OFPT10_FLOW_REMOVED) {
|
||
const struct ofp10_flow_removed *ofr;
|
||
|
||
ofr = ofpbuf_pull(&b, sizeof *ofr);
|
||
|
||
ofputil_match_from_ofp10_match(&ofr->match, &fr->match);
|
||
fr->priority = ntohs(ofr->priority);
|
||
fr->cookie = ofr->cookie;
|
||
fr->reason = ofr->reason;
|
||
fr->table_id = 255;
|
||
fr->duration_sec = ntohl(ofr->duration_sec);
|
||
fr->duration_nsec = ntohl(ofr->duration_nsec);
|
||
fr->idle_timeout = ntohs(ofr->idle_timeout);
|
||
fr->hard_timeout = 0;
|
||
fr->packet_count = ntohll(ofr->packet_count);
|
||
fr->byte_count = ntohll(ofr->byte_count);
|
||
} else if (raw == OFPRAW_NXT_FLOW_REMOVED) {
|
||
struct nx_flow_removed *nfr;
|
||
enum ofperr error;
|
||
|
||
nfr = ofpbuf_pull(&b, sizeof *nfr);
|
||
error = nx_pull_match(&b, ntohs(nfr->match_len), &fr->match,
|
||
NULL, NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
if (ofpbuf_size(&b)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
fr->priority = ntohs(nfr->priority);
|
||
fr->cookie = nfr->cookie;
|
||
fr->reason = nfr->reason;
|
||
fr->table_id = nfr->table_id ? nfr->table_id - 1 : 255;
|
||
fr->duration_sec = ntohl(nfr->duration_sec);
|
||
fr->duration_nsec = ntohl(nfr->duration_nsec);
|
||
fr->idle_timeout = ntohs(nfr->idle_timeout);
|
||
fr->hard_timeout = 0;
|
||
fr->packet_count = ntohll(nfr->packet_count);
|
||
fr->byte_count = ntohll(nfr->byte_count);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Converts abstract ofputil_flow_removed 'fr' into an OFPT_FLOW_REMOVED or
|
||
* NXT_FLOW_REMOVED message 'oh' according to 'protocol', and returns the
|
||
* message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_flow_removed(const struct ofputil_flow_removed *fr,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
struct ofpbuf *msg;
|
||
enum ofp_flow_removed_reason reason = fr->reason;
|
||
|
||
if (reason == OFPRR_METER_DELETE && !(protocol & OFPUTIL_P_OF14_UP)) {
|
||
reason = OFPRR_DELETE;
|
||
}
|
||
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF11_STD:
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM: {
|
||
struct ofp12_flow_removed *ofr;
|
||
|
||
msg = ofpraw_alloc_xid(OFPRAW_OFPT11_FLOW_REMOVED,
|
||
ofputil_protocol_to_ofp_version(protocol),
|
||
htonl(0),
|
||
ofputil_match_typical_len(protocol));
|
||
ofr = ofpbuf_put_zeros(msg, sizeof *ofr);
|
||
ofr->cookie = fr->cookie;
|
||
ofr->priority = htons(fr->priority);
|
||
ofr->reason = reason;
|
||
ofr->table_id = fr->table_id;
|
||
ofr->duration_sec = htonl(fr->duration_sec);
|
||
ofr->duration_nsec = htonl(fr->duration_nsec);
|
||
ofr->idle_timeout = htons(fr->idle_timeout);
|
||
ofr->hard_timeout = htons(fr->hard_timeout);
|
||
ofr->packet_count = htonll(fr->packet_count);
|
||
ofr->byte_count = htonll(fr->byte_count);
|
||
ofputil_put_ofp11_match(msg, &fr->match, protocol);
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID: {
|
||
struct ofp10_flow_removed *ofr;
|
||
|
||
msg = ofpraw_alloc_xid(OFPRAW_OFPT10_FLOW_REMOVED, OFP10_VERSION,
|
||
htonl(0), 0);
|
||
ofr = ofpbuf_put_zeros(msg, sizeof *ofr);
|
||
ofputil_match_to_ofp10_match(&fr->match, &ofr->match);
|
||
ofr->cookie = fr->cookie;
|
||
ofr->priority = htons(fr->priority);
|
||
ofr->reason = reason;
|
||
ofr->duration_sec = htonl(fr->duration_sec);
|
||
ofr->duration_nsec = htonl(fr->duration_nsec);
|
||
ofr->idle_timeout = htons(fr->idle_timeout);
|
||
ofr->packet_count = htonll(unknown_to_zero(fr->packet_count));
|
||
ofr->byte_count = htonll(unknown_to_zero(fr->byte_count));
|
||
break;
|
||
}
|
||
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID: {
|
||
struct nx_flow_removed *nfr;
|
||
int match_len;
|
||
|
||
msg = ofpraw_alloc_xid(OFPRAW_NXT_FLOW_REMOVED, OFP10_VERSION,
|
||
htonl(0), NXM_TYPICAL_LEN);
|
||
ofpbuf_put_zeros(msg, sizeof *nfr);
|
||
match_len = nx_put_match(msg, &fr->match, 0, 0);
|
||
|
||
nfr = ofpbuf_l3(msg);
|
||
nfr->cookie = fr->cookie;
|
||
nfr->priority = htons(fr->priority);
|
||
nfr->reason = reason;
|
||
nfr->table_id = fr->table_id + 1;
|
||
nfr->duration_sec = htonl(fr->duration_sec);
|
||
nfr->duration_nsec = htonl(fr->duration_nsec);
|
||
nfr->idle_timeout = htons(fr->idle_timeout);
|
||
nfr->match_len = htons(match_len);
|
||
nfr->packet_count = htonll(fr->packet_count);
|
||
nfr->byte_count = htonll(fr->byte_count);
|
||
break;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return msg;
|
||
}
|
||
|
||
static void
|
||
ofputil_decode_packet_in_finish(struct ofputil_packet_in *pin,
|
||
struct match *match, struct ofpbuf *b)
|
||
{
|
||
pin->packet = ofpbuf_data(b);
|
||
pin->packet_len = ofpbuf_size(b);
|
||
|
||
pin->fmd.in_port = match->flow.in_port.ofp_port;
|
||
pin->fmd.tun_id = match->flow.tunnel.tun_id;
|
||
pin->fmd.tun_src = match->flow.tunnel.ip_src;
|
||
pin->fmd.tun_dst = match->flow.tunnel.ip_dst;
|
||
pin->fmd.metadata = match->flow.metadata;
|
||
memcpy(pin->fmd.regs, match->flow.regs, sizeof pin->fmd.regs);
|
||
pin->fmd.pkt_mark = match->flow.pkt_mark;
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_packet_in(struct ofputil_packet_in *pin,
|
||
const struct ofp_header *oh)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
memset(pin, 0, sizeof *pin);
|
||
pin->cookie = OVS_BE64_MAX;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
if (raw == OFPRAW_OFPT13_PACKET_IN || raw == OFPRAW_OFPT12_PACKET_IN) {
|
||
const struct ofp13_packet_in *opi;
|
||
struct match match;
|
||
int error;
|
||
size_t packet_in_size;
|
||
|
||
if (raw == OFPRAW_OFPT12_PACKET_IN) {
|
||
packet_in_size = sizeof (struct ofp12_packet_in);
|
||
} else {
|
||
packet_in_size = sizeof (struct ofp13_packet_in);
|
||
}
|
||
|
||
opi = ofpbuf_pull(&b, packet_in_size);
|
||
error = oxm_pull_match_loose(&b, &match);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (!ofpbuf_try_pull(&b, 2)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
pin->reason = opi->pi.reason;
|
||
pin->table_id = opi->pi.table_id;
|
||
pin->buffer_id = ntohl(opi->pi.buffer_id);
|
||
pin->total_len = ntohs(opi->pi.total_len);
|
||
|
||
if (raw == OFPRAW_OFPT13_PACKET_IN) {
|
||
pin->cookie = opi->cookie;
|
||
}
|
||
|
||
ofputil_decode_packet_in_finish(pin, &match, &b);
|
||
} else if (raw == OFPRAW_OFPT10_PACKET_IN) {
|
||
const struct ofp10_packet_in *opi;
|
||
|
||
opi = ofpbuf_pull(&b, offsetof(struct ofp10_packet_in, data));
|
||
|
||
pin->packet = opi->data;
|
||
pin->packet_len = ofpbuf_size(&b);
|
||
|
||
pin->fmd.in_port = u16_to_ofp(ntohs(opi->in_port));
|
||
pin->reason = opi->reason;
|
||
pin->buffer_id = ntohl(opi->buffer_id);
|
||
pin->total_len = ntohs(opi->total_len);
|
||
} else if (raw == OFPRAW_OFPT11_PACKET_IN) {
|
||
const struct ofp11_packet_in *opi;
|
||
enum ofperr error;
|
||
|
||
opi = ofpbuf_pull(&b, sizeof *opi);
|
||
|
||
pin->packet = ofpbuf_data(&b);
|
||
pin->packet_len = ofpbuf_size(&b);
|
||
|
||
pin->buffer_id = ntohl(opi->buffer_id);
|
||
error = ofputil_port_from_ofp11(opi->in_port, &pin->fmd.in_port);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
pin->total_len = ntohs(opi->total_len);
|
||
pin->reason = opi->reason;
|
||
pin->table_id = opi->table_id;
|
||
} else if (raw == OFPRAW_NXT_PACKET_IN) {
|
||
const struct nx_packet_in *npi;
|
||
struct match match;
|
||
int error;
|
||
|
||
npi = ofpbuf_pull(&b, sizeof *npi);
|
||
error = nx_pull_match_loose(&b, ntohs(npi->match_len), &match, NULL,
|
||
NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (!ofpbuf_try_pull(&b, 2)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
pin->reason = npi->reason;
|
||
pin->table_id = npi->table_id;
|
||
pin->cookie = npi->cookie;
|
||
|
||
pin->buffer_id = ntohl(npi->buffer_id);
|
||
pin->total_len = ntohs(npi->total_len);
|
||
|
||
ofputil_decode_packet_in_finish(pin, &match, &b);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
static void
|
||
ofputil_packet_in_to_match(const struct ofputil_packet_in *pin,
|
||
struct match *match)
|
||
{
|
||
int i;
|
||
|
||
match_init_catchall(match);
|
||
if (pin->fmd.tun_id != htonll(0)) {
|
||
match_set_tun_id(match, pin->fmd.tun_id);
|
||
}
|
||
if (pin->fmd.tun_src != htonl(0)) {
|
||
match_set_tun_src(match, pin->fmd.tun_src);
|
||
}
|
||
if (pin->fmd.tun_dst != htonl(0)) {
|
||
match_set_tun_dst(match, pin->fmd.tun_dst);
|
||
}
|
||
if (pin->fmd.metadata != htonll(0)) {
|
||
match_set_metadata(match, pin->fmd.metadata);
|
||
}
|
||
|
||
for (i = 0; i < FLOW_N_REGS; i++) {
|
||
if (pin->fmd.regs[i]) {
|
||
match_set_reg(match, i, pin->fmd.regs[i]);
|
||
}
|
||
}
|
||
|
||
if (pin->fmd.pkt_mark != 0) {
|
||
match_set_pkt_mark(match, pin->fmd.pkt_mark);
|
||
}
|
||
|
||
match_set_in_port(match, pin->fmd.in_port);
|
||
}
|
||
|
||
static struct ofpbuf *
|
||
ofputil_encode_ofp10_packet_in(const struct ofputil_packet_in *pin)
|
||
{
|
||
struct ofp10_packet_in *opi;
|
||
struct ofpbuf *packet;
|
||
|
||
packet = ofpraw_alloc_xid(OFPRAW_OFPT10_PACKET_IN, OFP10_VERSION,
|
||
htonl(0), pin->packet_len);
|
||
opi = ofpbuf_put_zeros(packet, offsetof(struct ofp10_packet_in, data));
|
||
opi->total_len = htons(pin->total_len);
|
||
opi->in_port = htons(ofp_to_u16(pin->fmd.in_port));
|
||
opi->reason = pin->reason;
|
||
opi->buffer_id = htonl(pin->buffer_id);
|
||
|
||
ofpbuf_put(packet, pin->packet, pin->packet_len);
|
||
|
||
return packet;
|
||
}
|
||
|
||
static struct ofpbuf *
|
||
ofputil_encode_nx_packet_in(const struct ofputil_packet_in *pin)
|
||
{
|
||
struct nx_packet_in *npi;
|
||
struct ofpbuf *packet;
|
||
struct match match;
|
||
size_t match_len;
|
||
|
||
ofputil_packet_in_to_match(pin, &match);
|
||
|
||
/* The final argument is just an estimate of the space required. */
|
||
packet = ofpraw_alloc_xid(OFPRAW_NXT_PACKET_IN, OFP10_VERSION,
|
||
htonl(0), (sizeof(struct flow_metadata) * 2
|
||
+ 2 + pin->packet_len));
|
||
ofpbuf_put_zeros(packet, sizeof *npi);
|
||
match_len = nx_put_match(packet, &match, 0, 0);
|
||
ofpbuf_put_zeros(packet, 2);
|
||
ofpbuf_put(packet, pin->packet, pin->packet_len);
|
||
|
||
npi = ofpbuf_l3(packet);
|
||
npi->buffer_id = htonl(pin->buffer_id);
|
||
npi->total_len = htons(pin->total_len);
|
||
npi->reason = pin->reason;
|
||
npi->table_id = pin->table_id;
|
||
npi->cookie = pin->cookie;
|
||
npi->match_len = htons(match_len);
|
||
|
||
return packet;
|
||
}
|
||
|
||
static struct ofpbuf *
|
||
ofputil_encode_ofp11_packet_in(const struct ofputil_packet_in *pin)
|
||
{
|
||
struct ofp11_packet_in *opi;
|
||
struct ofpbuf *packet;
|
||
|
||
packet = ofpraw_alloc_xid(OFPRAW_OFPT11_PACKET_IN, OFP11_VERSION,
|
||
htonl(0), pin->packet_len);
|
||
opi = ofpbuf_put_zeros(packet, sizeof *opi);
|
||
opi->buffer_id = htonl(pin->buffer_id);
|
||
opi->in_port = ofputil_port_to_ofp11(pin->fmd.in_port);
|
||
opi->in_phy_port = opi->in_port;
|
||
opi->total_len = htons(pin->total_len);
|
||
opi->reason = pin->reason;
|
||
opi->table_id = pin->table_id;
|
||
|
||
ofpbuf_put(packet, pin->packet, pin->packet_len);
|
||
|
||
return packet;
|
||
}
|
||
|
||
static struct ofpbuf *
|
||
ofputil_encode_ofp12_packet_in(const struct ofputil_packet_in *pin,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
struct ofp13_packet_in *opi;
|
||
struct match match;
|
||
enum ofpraw packet_in_raw;
|
||
enum ofp_version packet_in_version;
|
||
size_t packet_in_size;
|
||
struct ofpbuf *packet;
|
||
|
||
if (protocol == OFPUTIL_P_OF12_OXM) {
|
||
packet_in_raw = OFPRAW_OFPT12_PACKET_IN;
|
||
packet_in_version = OFP12_VERSION;
|
||
packet_in_size = sizeof (struct ofp12_packet_in);
|
||
} else {
|
||
packet_in_raw = OFPRAW_OFPT13_PACKET_IN;
|
||
packet_in_version = OFP13_VERSION;
|
||
packet_in_size = sizeof (struct ofp13_packet_in);
|
||
}
|
||
|
||
ofputil_packet_in_to_match(pin, &match);
|
||
|
||
/* The final argument is just an estimate of the space required. */
|
||
packet = ofpraw_alloc_xid(packet_in_raw, packet_in_version,
|
||
htonl(0), (sizeof(struct flow_metadata) * 2
|
||
+ 2 + pin->packet_len));
|
||
ofpbuf_put_zeros(packet, packet_in_size);
|
||
oxm_put_match(packet, &match, ofputil_protocol_to_ofp_version(protocol));
|
||
ofpbuf_put_zeros(packet, 2);
|
||
ofpbuf_put(packet, pin->packet, pin->packet_len);
|
||
|
||
opi = ofpbuf_l3(packet);
|
||
opi->pi.buffer_id = htonl(pin->buffer_id);
|
||
opi->pi.total_len = htons(pin->total_len);
|
||
opi->pi.reason = pin->reason;
|
||
opi->pi.table_id = pin->table_id;
|
||
if (protocol == OFPUTIL_P_OF13_OXM) {
|
||
opi->cookie = pin->cookie;
|
||
}
|
||
|
||
return packet;
|
||
}
|
||
|
||
/* Converts abstract ofputil_packet_in 'pin' into a PACKET_IN message
|
||
* in the format specified by 'packet_in_format'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_packet_in(const struct ofputil_packet_in *pin,
|
||
enum ofputil_protocol protocol,
|
||
enum nx_packet_in_format packet_in_format)
|
||
{
|
||
struct ofpbuf *packet;
|
||
|
||
switch (protocol) {
|
||
case OFPUTIL_P_OF10_STD:
|
||
case OFPUTIL_P_OF10_STD_TID:
|
||
case OFPUTIL_P_OF10_NXM:
|
||
case OFPUTIL_P_OF10_NXM_TID:
|
||
packet = (packet_in_format == NXPIF_NXM
|
||
? ofputil_encode_nx_packet_in(pin)
|
||
: ofputil_encode_ofp10_packet_in(pin));
|
||
break;
|
||
|
||
case OFPUTIL_P_OF11_STD:
|
||
packet = ofputil_encode_ofp11_packet_in(pin);
|
||
break;
|
||
|
||
case OFPUTIL_P_OF12_OXM:
|
||
case OFPUTIL_P_OF13_OXM:
|
||
case OFPUTIL_P_OF14_OXM:
|
||
case OFPUTIL_P_OF15_OXM:
|
||
packet = ofputil_encode_ofp12_packet_in(pin, protocol);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
ofpmsg_update_length(packet);
|
||
return packet;
|
||
}
|
||
|
||
/* Returns a string form of 'reason'. The return value is either a statically
|
||
* allocated constant string or the 'bufsize'-byte buffer 'reasonbuf'.
|
||
* 'bufsize' should be at least OFPUTIL_PACKET_IN_REASON_BUFSIZE. */
|
||
const char *
|
||
ofputil_packet_in_reason_to_string(enum ofp_packet_in_reason reason,
|
||
char *reasonbuf, size_t bufsize)
|
||
{
|
||
switch (reason) {
|
||
case OFPR_NO_MATCH:
|
||
return "no_match";
|
||
case OFPR_ACTION:
|
||
return "action";
|
||
case OFPR_INVALID_TTL:
|
||
return "invalid_ttl";
|
||
|
||
case OFPR_N_REASONS:
|
||
default:
|
||
snprintf(reasonbuf, bufsize, "%d", (int) reason);
|
||
return reasonbuf;
|
||
}
|
||
}
|
||
|
||
bool
|
||
ofputil_packet_in_reason_from_string(const char *s,
|
||
enum ofp_packet_in_reason *reason)
|
||
{
|
||
int i;
|
||
|
||
for (i = 0; i < OFPR_N_REASONS; i++) {
|
||
char reasonbuf[OFPUTIL_PACKET_IN_REASON_BUFSIZE];
|
||
const char *reason_s;
|
||
|
||
reason_s = ofputil_packet_in_reason_to_string(i, reasonbuf,
|
||
sizeof reasonbuf);
|
||
if (!strcasecmp(s, reason_s)) {
|
||
*reason = i;
|
||
return true;
|
||
}
|
||
}
|
||
return false;
|
||
}
|
||
|
||
/* Converts an OFPT_PACKET_OUT in 'opo' into an abstract ofputil_packet_out in
|
||
* 'po'.
|
||
*
|
||
* Uses 'ofpacts' to store the abstract OFPACT_* version of the packet out
|
||
* message's actions. The caller must initialize 'ofpacts' and retains
|
||
* ownership of it. 'po->ofpacts' will point into the 'ofpacts' buffer.
|
||
*
|
||
* Returns 0 if successful, otherwise an OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_packet_out(struct ofputil_packet_out *po,
|
||
const struct ofp_header *oh,
|
||
struct ofpbuf *ofpacts)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
if (raw == OFPRAW_OFPT11_PACKET_OUT) {
|
||
enum ofperr error;
|
||
const struct ofp11_packet_out *opo = ofpbuf_pull(&b, sizeof *opo);
|
||
|
||
po->buffer_id = ntohl(opo->buffer_id);
|
||
error = ofputil_port_from_ofp11(opo->in_port, &po->in_port);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
error = ofpacts_pull_openflow_actions(&b, ntohs(opo->actions_len),
|
||
oh->version, ofpacts);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
} else if (raw == OFPRAW_OFPT10_PACKET_OUT) {
|
||
enum ofperr error;
|
||
const struct ofp10_packet_out *opo = ofpbuf_pull(&b, sizeof *opo);
|
||
|
||
po->buffer_id = ntohl(opo->buffer_id);
|
||
po->in_port = u16_to_ofp(ntohs(opo->in_port));
|
||
|
||
error = ofpacts_pull_openflow_actions(&b, ntohs(opo->actions_len),
|
||
oh->version, ofpacts);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
if (ofp_to_u16(po->in_port) >= ofp_to_u16(OFPP_MAX)
|
||
&& po->in_port != OFPP_LOCAL
|
||
&& po->in_port != OFPP_NONE && po->in_port != OFPP_CONTROLLER) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "packet-out has bad input port %#"PRIx16,
|
||
po->in_port);
|
||
return OFPERR_OFPBRC_BAD_PORT;
|
||
}
|
||
|
||
po->ofpacts = ofpbuf_data(ofpacts);
|
||
po->ofpacts_len = ofpbuf_size(ofpacts);
|
||
|
||
if (po->buffer_id == UINT32_MAX) {
|
||
po->packet = ofpbuf_data(&b);
|
||
po->packet_len = ofpbuf_size(&b);
|
||
} else {
|
||
po->packet = NULL;
|
||
po->packet_len = 0;
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* ofputil_phy_port */
|
||
|
||
/* NETDEV_F_* to and from OFPPF_* and OFPPF10_*. */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10MB_HD == OFPPF_10MB_HD); /* bit 0 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10MB_FD == OFPPF_10MB_FD); /* bit 1 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_100MB_HD == OFPPF_100MB_HD); /* bit 2 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_100MB_FD == OFPPF_100MB_FD); /* bit 3 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_1GB_HD == OFPPF_1GB_HD); /* bit 4 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_1GB_FD == OFPPF_1GB_FD); /* bit 5 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10GB_FD == OFPPF_10GB_FD); /* bit 6 */
|
||
|
||
/* NETDEV_F_ bits 11...15 are OFPPF10_ bits 7...11: */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_COPPER == (OFPPF10_COPPER << 4));
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_FIBER == (OFPPF10_FIBER << 4));
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_AUTONEG == (OFPPF10_AUTONEG << 4));
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_PAUSE == (OFPPF10_PAUSE << 4));
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_PAUSE_ASYM == (OFPPF10_PAUSE_ASYM << 4));
|
||
|
||
static enum netdev_features
|
||
netdev_port_features_from_ofp10(ovs_be32 ofp10_)
|
||
{
|
||
uint32_t ofp10 = ntohl(ofp10_);
|
||
return (ofp10 & 0x7f) | ((ofp10 & 0xf80) << 4);
|
||
}
|
||
|
||
static ovs_be32
|
||
netdev_port_features_to_ofp10(enum netdev_features features)
|
||
{
|
||
return htonl((features & 0x7f) | ((features & 0xf800) >> 4));
|
||
}
|
||
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10MB_HD == OFPPF_10MB_HD); /* bit 0 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10MB_FD == OFPPF_10MB_FD); /* bit 1 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_100MB_HD == OFPPF_100MB_HD); /* bit 2 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_100MB_FD == OFPPF_100MB_FD); /* bit 3 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_1GB_HD == OFPPF_1GB_HD); /* bit 4 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_1GB_FD == OFPPF_1GB_FD); /* bit 5 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_10GB_FD == OFPPF_10GB_FD); /* bit 6 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_40GB_FD == OFPPF11_40GB_FD); /* bit 7 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_100GB_FD == OFPPF11_100GB_FD); /* bit 8 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_1TB_FD == OFPPF11_1TB_FD); /* bit 9 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_OTHER == OFPPF11_OTHER); /* bit 10 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_COPPER == OFPPF11_COPPER); /* bit 11 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_FIBER == OFPPF11_FIBER); /* bit 12 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_AUTONEG == OFPPF11_AUTONEG); /* bit 13 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_PAUSE == OFPPF11_PAUSE); /* bit 14 */
|
||
BUILD_ASSERT_DECL((int) NETDEV_F_PAUSE_ASYM == OFPPF11_PAUSE_ASYM);/* bit 15 */
|
||
|
||
static enum netdev_features
|
||
netdev_port_features_from_ofp11(ovs_be32 ofp11)
|
||
{
|
||
return ntohl(ofp11) & 0xffff;
|
||
}
|
||
|
||
static ovs_be32
|
||
netdev_port_features_to_ofp11(enum netdev_features features)
|
||
{
|
||
return htonl(features & 0xffff);
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_decode_ofp10_phy_port(struct ofputil_phy_port *pp,
|
||
const struct ofp10_phy_port *opp)
|
||
{
|
||
pp->port_no = u16_to_ofp(ntohs(opp->port_no));
|
||
memcpy(pp->hw_addr, opp->hw_addr, OFP_ETH_ALEN);
|
||
ovs_strlcpy(pp->name, opp->name, OFP_MAX_PORT_NAME_LEN);
|
||
|
||
pp->config = ntohl(opp->config) & OFPPC10_ALL;
|
||
pp->state = ntohl(opp->state) & OFPPS10_ALL;
|
||
|
||
pp->curr = netdev_port_features_from_ofp10(opp->curr);
|
||
pp->advertised = netdev_port_features_from_ofp10(opp->advertised);
|
||
pp->supported = netdev_port_features_from_ofp10(opp->supported);
|
||
pp->peer = netdev_port_features_from_ofp10(opp->peer);
|
||
|
||
pp->curr_speed = netdev_features_to_bps(pp->curr, 0) / 1000;
|
||
pp->max_speed = netdev_features_to_bps(pp->supported, 0) / 1000;
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_decode_ofp11_port(struct ofputil_phy_port *pp,
|
||
const struct ofp11_port *op)
|
||
{
|
||
enum ofperr error;
|
||
|
||
error = ofputil_port_from_ofp11(op->port_no, &pp->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
memcpy(pp->hw_addr, op->hw_addr, OFP_ETH_ALEN);
|
||
ovs_strlcpy(pp->name, op->name, OFP_MAX_PORT_NAME_LEN);
|
||
|
||
pp->config = ntohl(op->config) & OFPPC11_ALL;
|
||
pp->state = ntohl(op->state) & OFPPS11_ALL;
|
||
|
||
pp->curr = netdev_port_features_from_ofp11(op->curr);
|
||
pp->advertised = netdev_port_features_from_ofp11(op->advertised);
|
||
pp->supported = netdev_port_features_from_ofp11(op->supported);
|
||
pp->peer = netdev_port_features_from_ofp11(op->peer);
|
||
|
||
pp->curr_speed = ntohl(op->curr_speed);
|
||
pp->max_speed = ntohl(op->max_speed);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_ofp14_port_ethernet_property(const struct ofpbuf *payload,
|
||
struct ofputil_phy_port *pp)
|
||
{
|
||
struct ofp14_port_desc_prop_ethernet *eth = ofpbuf_data(payload);
|
||
|
||
if (ofpbuf_size(payload) != sizeof *eth) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
|
||
pp->curr = netdev_port_features_from_ofp11(eth->curr);
|
||
pp->advertised = netdev_port_features_from_ofp11(eth->advertised);
|
||
pp->supported = netdev_port_features_from_ofp11(eth->supported);
|
||
pp->peer = netdev_port_features_from_ofp11(eth->peer);
|
||
|
||
pp->curr_speed = ntohl(eth->curr_speed);
|
||
pp->max_speed = ntohl(eth->max_speed);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_ofp14_port(struct ofputil_phy_port *pp, struct ofpbuf *msg)
|
||
{
|
||
struct ofpbuf properties;
|
||
struct ofp14_port *op;
|
||
enum ofperr error;
|
||
size_t len;
|
||
|
||
op = ofpbuf_try_pull(msg, sizeof *op);
|
||
if (!op) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
len = ntohs(op->length);
|
||
if (len < sizeof *op || len - sizeof *op > ofpbuf_size(msg)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
len -= sizeof *op;
|
||
ofpbuf_use_const(&properties, ofpbuf_pull(msg, len), len);
|
||
|
||
error = ofputil_port_from_ofp11(op->port_no, &pp->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
memcpy(pp->hw_addr, op->hw_addr, OFP_ETH_ALEN);
|
||
ovs_strlcpy(pp->name, op->name, OFP_MAX_PORT_NAME_LEN);
|
||
|
||
pp->config = ntohl(op->config) & OFPPC11_ALL;
|
||
pp->state = ntohl(op->state) & OFPPS11_ALL;
|
||
|
||
while (ofpbuf_size(&properties) > 0) {
|
||
struct ofpbuf payload;
|
||
enum ofperr error;
|
||
uint16_t type;
|
||
|
||
error = ofputil_pull_property(&properties, &payload, &type);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
switch (type) {
|
||
case OFPPDPT14_ETHERNET:
|
||
error = parse_ofp14_port_ethernet_property(&payload, pp);
|
||
break;
|
||
|
||
default:
|
||
log_property(true, "unknown port property %"PRIu16, type);
|
||
error = 0;
|
||
break;
|
||
}
|
||
|
||
if (error) {
|
||
return error;
|
||
}
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
static void
|
||
ofputil_encode_ofp10_phy_port(const struct ofputil_phy_port *pp,
|
||
struct ofp10_phy_port *opp)
|
||
{
|
||
memset(opp, 0, sizeof *opp);
|
||
|
||
opp->port_no = htons(ofp_to_u16(pp->port_no));
|
||
memcpy(opp->hw_addr, pp->hw_addr, ETH_ADDR_LEN);
|
||
ovs_strlcpy(opp->name, pp->name, OFP_MAX_PORT_NAME_LEN);
|
||
|
||
opp->config = htonl(pp->config & OFPPC10_ALL);
|
||
opp->state = htonl(pp->state & OFPPS10_ALL);
|
||
|
||
opp->curr = netdev_port_features_to_ofp10(pp->curr);
|
||
opp->advertised = netdev_port_features_to_ofp10(pp->advertised);
|
||
opp->supported = netdev_port_features_to_ofp10(pp->supported);
|
||
opp->peer = netdev_port_features_to_ofp10(pp->peer);
|
||
}
|
||
|
||
static void
|
||
ofputil_encode_ofp11_port(const struct ofputil_phy_port *pp,
|
||
struct ofp11_port *op)
|
||
{
|
||
memset(op, 0, sizeof *op);
|
||
|
||
op->port_no = ofputil_port_to_ofp11(pp->port_no);
|
||
memcpy(op->hw_addr, pp->hw_addr, ETH_ADDR_LEN);
|
||
ovs_strlcpy(op->name, pp->name, OFP_MAX_PORT_NAME_LEN);
|
||
|
||
op->config = htonl(pp->config & OFPPC11_ALL);
|
||
op->state = htonl(pp->state & OFPPS11_ALL);
|
||
|
||
op->curr = netdev_port_features_to_ofp11(pp->curr);
|
||
op->advertised = netdev_port_features_to_ofp11(pp->advertised);
|
||
op->supported = netdev_port_features_to_ofp11(pp->supported);
|
||
op->peer = netdev_port_features_to_ofp11(pp->peer);
|
||
|
||
op->curr_speed = htonl(pp->curr_speed);
|
||
op->max_speed = htonl(pp->max_speed);
|
||
}
|
||
|
||
static void
|
||
ofputil_put_ofp14_port(const struct ofputil_phy_port *pp,
|
||
struct ofpbuf *b)
|
||
{
|
||
struct ofp14_port *op;
|
||
struct ofp14_port_desc_prop_ethernet *eth;
|
||
|
||
ofpbuf_prealloc_tailroom(b, sizeof *op + sizeof *eth);
|
||
|
||
op = ofpbuf_put_zeros(b, sizeof *op);
|
||
op->port_no = ofputil_port_to_ofp11(pp->port_no);
|
||
op->length = htons(sizeof *op + sizeof *eth);
|
||
memcpy(op->hw_addr, pp->hw_addr, ETH_ADDR_LEN);
|
||
ovs_strlcpy(op->name, pp->name, sizeof op->name);
|
||
op->config = htonl(pp->config & OFPPC11_ALL);
|
||
op->state = htonl(pp->state & OFPPS11_ALL);
|
||
|
||
eth = ofpbuf_put_zeros(b, sizeof *eth);
|
||
eth->type = htons(OFPPDPT14_ETHERNET);
|
||
eth->length = htons(sizeof *eth);
|
||
eth->curr = netdev_port_features_to_ofp11(pp->curr);
|
||
eth->advertised = netdev_port_features_to_ofp11(pp->advertised);
|
||
eth->supported = netdev_port_features_to_ofp11(pp->supported);
|
||
eth->peer = netdev_port_features_to_ofp11(pp->peer);
|
||
eth->curr_speed = htonl(pp->curr_speed);
|
||
eth->max_speed = htonl(pp->max_speed);
|
||
}
|
||
|
||
static void
|
||
ofputil_put_phy_port(enum ofp_version ofp_version,
|
||
const struct ofputil_phy_port *pp, struct ofpbuf *b)
|
||
{
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
struct ofp10_phy_port *opp = ofpbuf_put_uninit(b, sizeof *opp);
|
||
ofputil_encode_ofp10_phy_port(pp, opp);
|
||
break;
|
||
}
|
||
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION: {
|
||
struct ofp11_port *op = ofpbuf_put_uninit(b, sizeof *op);
|
||
ofputil_encode_ofp11_port(pp, op);
|
||
break;
|
||
}
|
||
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
ofputil_put_ofp14_port(pp, b);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_port_desc_stats_request(const struct ofp_header *request,
|
||
ofp_port_t *port)
|
||
{
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
|
||
ofpbuf_use_const(&b, request, ntohs(request->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
if (raw == OFPRAW_OFPST10_PORT_DESC_REQUEST) {
|
||
*port = OFPP_ANY;
|
||
return 0;
|
||
} else if (raw == OFPRAW_OFPST15_PORT_DESC_REQUEST) {
|
||
ovs_be32 *ofp11_port;
|
||
|
||
ofp11_port = ofpbuf_pull(&b, sizeof *ofp11_port);
|
||
return ofputil_port_from_ofp11(*ofp11_port, port);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_port_desc_stats_request(enum ofp_version ofp_version,
|
||
ofp_port_t port)
|
||
{
|
||
struct ofpbuf *request;
|
||
ovs_be32 ofp11_port;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST10_PORT_DESC_REQUEST,
|
||
ofp_version, 0);
|
||
break;
|
||
|
||
case OFP15_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST15_PORT_DESC_REQUEST,
|
||
ofp_version, 0);
|
||
ofp11_port = ofputil_port_to_ofp11(port);
|
||
ofpbuf_put(request, &ofp11_port, sizeof ofp11_port);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
void
|
||
ofputil_append_port_desc_stats_reply(const struct ofputil_phy_port *pp,
|
||
struct list *replies)
|
||
{
|
||
struct ofpbuf *reply = ofpbuf_from_list(list_back(replies));
|
||
size_t start_ofs = ofpbuf_size(reply);
|
||
|
||
ofputil_put_phy_port(ofpmp_version(replies), pp, reply);
|
||
ofpmp_postappend(replies, start_ofs);
|
||
}
|
||
|
||
/* ofputil_switch_features */
|
||
|
||
#define OFPC_COMMON (OFPC_FLOW_STATS | OFPC_TABLE_STATS | OFPC_PORT_STATS | \
|
||
OFPC_IP_REASM | OFPC_QUEUE_STATS)
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_FLOW_STATS == OFPC_FLOW_STATS);
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_TABLE_STATS == OFPC_TABLE_STATS);
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_PORT_STATS == OFPC_PORT_STATS);
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_IP_REASM == OFPC_IP_REASM);
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_QUEUE_STATS == OFPC_QUEUE_STATS);
|
||
BUILD_ASSERT_DECL((int) OFPUTIL_C_ARP_MATCH_IP == OFPC_ARP_MATCH_IP);
|
||
|
||
static uint32_t
|
||
ofputil_capabilities_mask(enum ofp_version ofp_version)
|
||
{
|
||
/* Handle capabilities whose bit is unique for all Open Flow versions */
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
case OFP11_VERSION:
|
||
return OFPC_COMMON | OFPC_ARP_MATCH_IP;
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
return OFPC_COMMON | OFPC12_PORT_BLOCKED;
|
||
default:
|
||
/* Caller needs to check osf->header.version itself */
|
||
return 0;
|
||
}
|
||
}
|
||
|
||
/* Decodes an OpenFlow 1.0 or 1.1 "switch_features" structure 'osf' into an
|
||
* abstract representation in '*features'. Initializes '*b' to iterate over
|
||
* the OpenFlow port structures following 'osf' with later calls to
|
||
* ofputil_pull_phy_port(). Returns 0 if successful, otherwise an
|
||
* OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_switch_features(const struct ofp_header *oh,
|
||
struct ofputil_switch_features *features,
|
||
struct ofpbuf *b)
|
||
{
|
||
const struct ofp_switch_features *osf;
|
||
enum ofpraw raw;
|
||
|
||
ofpbuf_use_const(b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(b);
|
||
|
||
osf = ofpbuf_pull(b, sizeof *osf);
|
||
features->datapath_id = ntohll(osf->datapath_id);
|
||
features->n_buffers = ntohl(osf->n_buffers);
|
||
features->n_tables = osf->n_tables;
|
||
features->auxiliary_id = 0;
|
||
|
||
features->capabilities = ntohl(osf->capabilities) &
|
||
ofputil_capabilities_mask(oh->version);
|
||
|
||
if (raw == OFPRAW_OFPT10_FEATURES_REPLY) {
|
||
if (osf->capabilities & htonl(OFPC10_STP)) {
|
||
features->capabilities |= OFPUTIL_C_STP;
|
||
}
|
||
features->ofpacts = ofpact_bitmap_from_openflow(osf->actions,
|
||
OFP10_VERSION);
|
||
} else if (raw == OFPRAW_OFPT11_FEATURES_REPLY
|
||
|| raw == OFPRAW_OFPT13_FEATURES_REPLY) {
|
||
if (osf->capabilities & htonl(OFPC11_GROUP_STATS)) {
|
||
features->capabilities |= OFPUTIL_C_GROUP_STATS;
|
||
}
|
||
features->ofpacts = 0;
|
||
if (raw == OFPRAW_OFPT13_FEATURES_REPLY) {
|
||
features->auxiliary_id = osf->auxiliary_id;
|
||
}
|
||
} else {
|
||
return OFPERR_OFPBRC_BAD_VERSION;
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* In OpenFlow 1.0, 1.1, and 1.2, an OFPT_FEATURES_REPLY message lists all the
|
||
* switch's ports, unless there are too many to fit. In OpenFlow 1.3 and
|
||
* later, an OFPT_FEATURES_REPLY does not list ports at all.
|
||
*
|
||
* Given a buffer 'b' that contains a Features Reply message, this message
|
||
* checks if it contains a complete list of the switch's ports. Returns true,
|
||
* if so. Returns false if the list is missing (OF1.3+) or incomplete
|
||
* (OF1.0/1.1/1.2), and in the latter case removes all of the ports from the
|
||
* message.
|
||
*
|
||
* When this function returns false, the caller should send an OFPST_PORT_DESC
|
||
* stats request to get the ports. */
|
||
bool
|
||
ofputil_switch_features_has_ports(struct ofpbuf *b)
|
||
{
|
||
struct ofp_header *oh = ofpbuf_data(b);
|
||
size_t phy_port_size;
|
||
|
||
if (oh->version >= OFP13_VERSION) {
|
||
/* OpenFlow 1.3+ never has ports in the feature reply. */
|
||
return false;
|
||
}
|
||
|
||
phy_port_size = (oh->version == OFP10_VERSION
|
||
? sizeof(struct ofp10_phy_port)
|
||
: sizeof(struct ofp11_port));
|
||
if (ntohs(oh->length) + phy_port_size <= UINT16_MAX) {
|
||
/* There's room for additional ports in the feature reply.
|
||
* Assume that the list is complete. */
|
||
return true;
|
||
}
|
||
|
||
/* The feature reply has no room for more ports. Probably the list is
|
||
* truncated. Drop the ports and tell the caller to retrieve them with
|
||
* OFPST_PORT_DESC. */
|
||
ofpbuf_set_size(b, sizeof *oh + sizeof(struct ofp_switch_features));
|
||
ofpmsg_update_length(b);
|
||
return false;
|
||
}
|
||
|
||
/* Returns a buffer owned by the caller that encodes 'features' in the format
|
||
* required by 'protocol' with the given 'xid'. The caller should append port
|
||
* information to the buffer with subsequent calls to
|
||
* ofputil_put_switch_features_port(). */
|
||
struct ofpbuf *
|
||
ofputil_encode_switch_features(const struct ofputil_switch_features *features,
|
||
enum ofputil_protocol protocol, ovs_be32 xid)
|
||
{
|
||
struct ofp_switch_features *osf;
|
||
struct ofpbuf *b;
|
||
enum ofp_version version;
|
||
enum ofpraw raw;
|
||
|
||
version = ofputil_protocol_to_ofp_version(protocol);
|
||
switch (version) {
|
||
case OFP10_VERSION:
|
||
raw = OFPRAW_OFPT10_FEATURES_REPLY;
|
||
break;
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
raw = OFPRAW_OFPT11_FEATURES_REPLY;
|
||
break;
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
raw = OFPRAW_OFPT13_FEATURES_REPLY;
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
b = ofpraw_alloc_xid(raw, version, xid, 0);
|
||
osf = ofpbuf_put_zeros(b, sizeof *osf);
|
||
osf->datapath_id = htonll(features->datapath_id);
|
||
osf->n_buffers = htonl(features->n_buffers);
|
||
osf->n_tables = features->n_tables;
|
||
|
||
osf->capabilities = htonl(features->capabilities & OFPC_COMMON);
|
||
osf->capabilities = htonl(features->capabilities &
|
||
ofputil_capabilities_mask(version));
|
||
switch (version) {
|
||
case OFP10_VERSION:
|
||
if (features->capabilities & OFPUTIL_C_STP) {
|
||
osf->capabilities |= htonl(OFPC10_STP);
|
||
}
|
||
osf->actions = ofpact_bitmap_to_openflow(features->ofpacts,
|
||
OFP10_VERSION);
|
||
break;
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
osf->auxiliary_id = features->auxiliary_id;
|
||
/* fall through */
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
if (features->capabilities & OFPUTIL_C_GROUP_STATS) {
|
||
osf->capabilities |= htonl(OFPC11_GROUP_STATS);
|
||
}
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return b;
|
||
}
|
||
|
||
/* Encodes 'pp' into the format required by the switch_features message already
|
||
* in 'b', which should have been returned by ofputil_encode_switch_features(),
|
||
* and appends the encoded version to 'b'. */
|
||
void
|
||
ofputil_put_switch_features_port(const struct ofputil_phy_port *pp,
|
||
struct ofpbuf *b)
|
||
{
|
||
const struct ofp_header *oh = ofpbuf_data(b);
|
||
|
||
if (oh->version < OFP13_VERSION) {
|
||
/* Try adding a port description to the message, but drop it again if
|
||
* the buffer overflows. (This possibility for overflow is why
|
||
* OpenFlow 1.3+ moved port descriptions into a multipart message.) */
|
||
size_t start_ofs = ofpbuf_size(b);
|
||
ofputil_put_phy_port(oh->version, pp, b);
|
||
if (ofpbuf_size(b) > UINT16_MAX) {
|
||
ofpbuf_set_size(b, start_ofs);
|
||
}
|
||
}
|
||
}
|
||
|
||
/* ofputil_port_status */
|
||
|
||
/* Decodes the OpenFlow "port status" message in '*ops' into an abstract form
|
||
* in '*ps'. Returns 0 if successful, otherwise an OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_port_status(const struct ofp_header *oh,
|
||
struct ofputil_port_status *ps)
|
||
{
|
||
const struct ofp_port_status *ops;
|
||
struct ofpbuf b;
|
||
int retval;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
ofpraw_pull_assert(&b);
|
||
ops = ofpbuf_pull(&b, sizeof *ops);
|
||
|
||
if (ops->reason != OFPPR_ADD &&
|
||
ops->reason != OFPPR_DELETE &&
|
||
ops->reason != OFPPR_MODIFY) {
|
||
return OFPERR_NXBRC_BAD_REASON;
|
||
}
|
||
ps->reason = ops->reason;
|
||
|
||
retval = ofputil_pull_phy_port(oh->version, &b, &ps->desc);
|
||
ovs_assert(retval != EOF);
|
||
return retval;
|
||
}
|
||
|
||
/* Converts the abstract form of a "port status" message in '*ps' into an
|
||
* OpenFlow message suitable for 'protocol', and returns that encoded form in
|
||
* a buffer owned by the caller. */
|
||
struct ofpbuf *
|
||
ofputil_encode_port_status(const struct ofputil_port_status *ps,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
struct ofp_port_status *ops;
|
||
struct ofpbuf *b;
|
||
enum ofp_version version;
|
||
enum ofpraw raw;
|
||
|
||
version = ofputil_protocol_to_ofp_version(protocol);
|
||
switch (version) {
|
||
case OFP10_VERSION:
|
||
raw = OFPRAW_OFPT10_PORT_STATUS;
|
||
break;
|
||
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
raw = OFPRAW_OFPT11_PORT_STATUS;
|
||
break;
|
||
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
raw = OFPRAW_OFPT14_PORT_STATUS;
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
b = ofpraw_alloc_xid(raw, version, htonl(0), 0);
|
||
ops = ofpbuf_put_zeros(b, sizeof *ops);
|
||
ops->reason = ps->reason;
|
||
ofputil_put_phy_port(version, &ps->desc, b);
|
||
ofpmsg_update_length(b);
|
||
return b;
|
||
}
|
||
|
||
/* ofputil_port_mod */
|
||
|
||
static enum ofperr
|
||
parse_port_mod_ethernet_property(struct ofpbuf *property,
|
||
struct ofputil_port_mod *pm)
|
||
{
|
||
struct ofp14_port_mod_prop_ethernet *eth = ofpbuf_data(property);
|
||
|
||
if (ofpbuf_size(property) != sizeof *eth) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
pm->advertise = netdev_port_features_from_ofp11(eth->advertise);
|
||
return 0;
|
||
}
|
||
|
||
/* Decodes the OpenFlow "port mod" message in '*oh' into an abstract form in
|
||
* '*pm'. Returns 0 if successful, otherwise an OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_port_mod(const struct ofp_header *oh,
|
||
struct ofputil_port_mod *pm, bool loose)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
if (raw == OFPRAW_OFPT10_PORT_MOD) {
|
||
const struct ofp10_port_mod *opm = ofpbuf_data(&b);
|
||
|
||
pm->port_no = u16_to_ofp(ntohs(opm->port_no));
|
||
memcpy(pm->hw_addr, opm->hw_addr, ETH_ADDR_LEN);
|
||
pm->config = ntohl(opm->config) & OFPPC10_ALL;
|
||
pm->mask = ntohl(opm->mask) & OFPPC10_ALL;
|
||
pm->advertise = netdev_port_features_from_ofp10(opm->advertise);
|
||
} else if (raw == OFPRAW_OFPT11_PORT_MOD) {
|
||
const struct ofp11_port_mod *opm = ofpbuf_data(&b);
|
||
enum ofperr error;
|
||
|
||
error = ofputil_port_from_ofp11(opm->port_no, &pm->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
memcpy(pm->hw_addr, opm->hw_addr, ETH_ADDR_LEN);
|
||
pm->config = ntohl(opm->config) & OFPPC11_ALL;
|
||
pm->mask = ntohl(opm->mask) & OFPPC11_ALL;
|
||
pm->advertise = netdev_port_features_from_ofp11(opm->advertise);
|
||
} else if (raw == OFPRAW_OFPT14_PORT_MOD) {
|
||
const struct ofp14_port_mod *opm = ofpbuf_pull(&b, sizeof *opm);
|
||
enum ofperr error;
|
||
|
||
memset(pm, 0, sizeof *pm);
|
||
|
||
error = ofputil_port_from_ofp11(opm->port_no, &pm->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
memcpy(pm->hw_addr, opm->hw_addr, ETH_ADDR_LEN);
|
||
pm->config = ntohl(opm->config) & OFPPC11_ALL;
|
||
pm->mask = ntohl(opm->mask) & OFPPC11_ALL;
|
||
|
||
while (ofpbuf_size(&b) > 0) {
|
||
struct ofpbuf property;
|
||
enum ofperr error;
|
||
uint16_t type;
|
||
|
||
error = ofputil_pull_property(&b, &property, &type);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
switch (type) {
|
||
case OFPPMPT14_ETHERNET:
|
||
error = parse_port_mod_ethernet_property(&property, pm);
|
||
break;
|
||
|
||
default:
|
||
log_property(loose, "unknown port_mod property %"PRIu16, type);
|
||
if (loose) {
|
||
error = 0;
|
||
} else if (type == OFPPMPT14_EXPERIMENTER) {
|
||
error = OFPERR_OFPBPC_BAD_EXPERIMENTER;
|
||
} else {
|
||
error = OFPERR_OFPBRC_BAD_TYPE;
|
||
}
|
||
break;
|
||
}
|
||
|
||
if (error) {
|
||
return error;
|
||
}
|
||
}
|
||
} else {
|
||
return OFPERR_OFPBRC_BAD_TYPE;
|
||
}
|
||
|
||
pm->config &= pm->mask;
|
||
return 0;
|
||
}
|
||
|
||
/* Converts the abstract form of a "port mod" message in '*pm' into an OpenFlow
|
||
* message suitable for 'protocol', and returns that encoded form in a buffer
|
||
* owned by the caller. */
|
||
struct ofpbuf *
|
||
ofputil_encode_port_mod(const struct ofputil_port_mod *pm,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
enum ofp_version ofp_version = ofputil_protocol_to_ofp_version(protocol);
|
||
struct ofpbuf *b;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
struct ofp10_port_mod *opm;
|
||
|
||
b = ofpraw_alloc(OFPRAW_OFPT10_PORT_MOD, ofp_version, 0);
|
||
opm = ofpbuf_put_zeros(b, sizeof *opm);
|
||
opm->port_no = htons(ofp_to_u16(pm->port_no));
|
||
memcpy(opm->hw_addr, pm->hw_addr, ETH_ADDR_LEN);
|
||
opm->config = htonl(pm->config & OFPPC10_ALL);
|
||
opm->mask = htonl(pm->mask & OFPPC10_ALL);
|
||
opm->advertise = netdev_port_features_to_ofp10(pm->advertise);
|
||
break;
|
||
}
|
||
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION: {
|
||
struct ofp11_port_mod *opm;
|
||
|
||
b = ofpraw_alloc(OFPRAW_OFPT11_PORT_MOD, ofp_version, 0);
|
||
opm = ofpbuf_put_zeros(b, sizeof *opm);
|
||
opm->port_no = ofputil_port_to_ofp11(pm->port_no);
|
||
memcpy(opm->hw_addr, pm->hw_addr, ETH_ADDR_LEN);
|
||
opm->config = htonl(pm->config & OFPPC11_ALL);
|
||
opm->mask = htonl(pm->mask & OFPPC11_ALL);
|
||
opm->advertise = netdev_port_features_to_ofp11(pm->advertise);
|
||
break;
|
||
}
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp14_port_mod_prop_ethernet *eth;
|
||
struct ofp14_port_mod *opm;
|
||
|
||
b = ofpraw_alloc(OFPRAW_OFPT14_PORT_MOD, ofp_version, sizeof *eth);
|
||
opm = ofpbuf_put_zeros(b, sizeof *opm);
|
||
opm->port_no = ofputil_port_to_ofp11(pm->port_no);
|
||
memcpy(opm->hw_addr, pm->hw_addr, ETH_ADDR_LEN);
|
||
opm->config = htonl(pm->config & OFPPC11_ALL);
|
||
opm->mask = htonl(pm->mask & OFPPC11_ALL);
|
||
|
||
if (pm->advertise) {
|
||
eth = ofpbuf_put_zeros(b, sizeof *eth);
|
||
eth->type = htons(OFPPMPT14_ETHERNET);
|
||
eth->length = htons(sizeof *eth);
|
||
eth->advertise = netdev_port_features_to_ofp11(pm->advertise);
|
||
}
|
||
break;
|
||
}
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return b;
|
||
}
|
||
|
||
/* Table features. */
|
||
|
||
static enum ofperr
|
||
pull_table_feature_property(struct ofpbuf *msg, struct ofpbuf *payload,
|
||
uint16_t *typep)
|
||
{
|
||
enum ofperr error;
|
||
|
||
error = ofputil_pull_property(msg, payload, typep);
|
||
if (payload && !error) {
|
||
ofpbuf_pull(payload, sizeof(struct ofp_prop_header));
|
||
}
|
||
return error;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_action_bitmap(struct ofpbuf *payload, enum ofp_version ofp_version,
|
||
uint64_t *ofpacts)
|
||
{
|
||
uint32_t types = 0;
|
||
|
||
while (ofpbuf_size(payload) > 0) {
|
||
uint16_t type;
|
||
enum ofperr error;
|
||
|
||
error = ofputil_pull_property__(payload, NULL, 1, &type);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
if (type < CHAR_BIT * sizeof types) {
|
||
types |= 1u << type;
|
||
}
|
||
}
|
||
|
||
*ofpacts = ofpact_bitmap_from_openflow(htonl(types), ofp_version);
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_instruction_ids(struct ofpbuf *payload, bool loose, uint32_t *insts)
|
||
{
|
||
*insts = 0;
|
||
while (ofpbuf_size(payload) > 0) {
|
||
enum ovs_instruction_type inst;
|
||
enum ofperr error;
|
||
uint16_t ofpit;
|
||
|
||
/* OF1.3 and OF1.4 aren't clear about padding in the instruction IDs.
|
||
* It seems clear that they aren't padded to 8 bytes, though, because
|
||
* both standards say that "non-experimenter instructions are 4 bytes"
|
||
* and do not mention any padding before the first instruction ID.
|
||
* (There wouldn't be any point in padding to 8 bytes if the IDs were
|
||
* aligned on an odd 4-byte boundary.)
|
||
*
|
||
* Anyway, we just assume they're all glommed together on byte
|
||
* boundaries. */
|
||
error = ofputil_pull_property__(payload, NULL, 1, &ofpit);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
error = ovs_instruction_type_from_inst_type(&inst, ofpit);
|
||
if (!error) {
|
||
*insts |= 1u << inst;
|
||
} else if (!loose) {
|
||
return error;
|
||
}
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_table_features_next_table(struct ofpbuf *payload,
|
||
unsigned long int *next_tables)
|
||
{
|
||
size_t i;
|
||
|
||
memset(next_tables, 0, bitmap_n_bytes(255));
|
||
for (i = 0; i < ofpbuf_size(payload); i++) {
|
||
uint8_t id = ((const uint8_t *) ofpbuf_data(payload))[i];
|
||
if (id >= 255) {
|
||
return OFPERR_OFPBPC_BAD_VALUE;
|
||
}
|
||
bitmap_set1(next_tables, id);
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_oxm(struct ofpbuf *b, bool loose,
|
||
const struct mf_field **fieldp, bool *hasmask)
|
||
{
|
||
ovs_be32 *oxmp;
|
||
uint32_t oxm;
|
||
|
||
oxmp = ofpbuf_try_pull(b, sizeof *oxmp);
|
||
if (!oxmp) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
oxm = ntohl(*oxmp);
|
||
|
||
/* Determine '*hasmask'. If 'oxm' is masked, convert it to the equivalent
|
||
* unmasked version, because the table of OXM fields we support only has
|
||
* masked versions of fields that we support with masks, but we should be
|
||
* able to parse the masked versions of those here. */
|
||
*hasmask = NXM_HASMASK(oxm);
|
||
if (*hasmask) {
|
||
if (NXM_LENGTH(oxm) & 1) {
|
||
return OFPERR_OFPBPC_BAD_VALUE;
|
||
}
|
||
oxm = NXM_HEADER(NXM_VENDOR(oxm), NXM_FIELD(oxm), NXM_LENGTH(oxm) / 2);
|
||
}
|
||
|
||
*fieldp = mf_from_nxm_header(oxm);
|
||
if (!*fieldp) {
|
||
log_property(loose, "unknown OXM field %#"PRIx32, ntohl(*oxmp));
|
||
}
|
||
return *fieldp ? 0 : OFPERR_OFPBMC_BAD_FIELD;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_oxms(struct ofpbuf *payload, bool loose,
|
||
struct mf_bitmap *exactp, struct mf_bitmap *maskedp)
|
||
{
|
||
struct mf_bitmap exact = MF_BITMAP_INITIALIZER;
|
||
struct mf_bitmap masked = MF_BITMAP_INITIALIZER;
|
||
|
||
while (ofpbuf_size(payload) > 0) {
|
||
const struct mf_field *field;
|
||
enum ofperr error;
|
||
bool hasmask;
|
||
|
||
error = parse_oxm(payload, loose, &field, &hasmask);
|
||
if (!error) {
|
||
bitmap_set1(hasmask ? masked.bm : exact.bm, field->id);
|
||
} else if (error != OFPERR_OFPBMC_BAD_FIELD || !loose) {
|
||
return error;
|
||
}
|
||
}
|
||
if (exactp) {
|
||
*exactp = exact;
|
||
} else if (!bitmap_is_all_zeros(exact.bm, MFF_N_IDS)) {
|
||
return OFPERR_OFPBMC_BAD_MASK;
|
||
}
|
||
if (maskedp) {
|
||
*maskedp = masked;
|
||
} else if (!bitmap_is_all_zeros(masked.bm, MFF_N_IDS)) {
|
||
return OFPERR_OFPBMC_BAD_MASK;
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
/* Converts an OFPMP_TABLE_FEATURES request or reply in 'msg' into an abstract
|
||
* ofputil_table_features in 'tf'.
|
||
*
|
||
* If 'loose' is true, this function ignores properties and values that it does
|
||
* not understand, as a controller would want to do when interpreting
|
||
* capabilities provided by a switch. If 'loose' is false, this function
|
||
* treats unknown properties and values as an error, as a switch would want to
|
||
* do when interpreting a configuration request made by a controller.
|
||
*
|
||
* A single OpenFlow message can specify features for multiple tables. Calling
|
||
* this function multiple times for a single 'msg' iterates through the tables
|
||
* in the message. The caller must initially leave 'msg''s layer pointers null
|
||
* and not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no tables were left in this 'msg', otherwise
|
||
* a positive "enum ofperr" value. */
|
||
int
|
||
ofputil_decode_table_features(struct ofpbuf *msg,
|
||
struct ofputil_table_features *tf, bool loose)
|
||
{
|
||
const struct ofp_header *oh;
|
||
struct ofp13_table_features *otf;
|
||
struct ofpbuf properties;
|
||
unsigned int len;
|
||
|
||
memset(tf, 0, sizeof *tf);
|
||
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
oh = ofpbuf_l2(msg);
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
if (ofpbuf_size(msg) < sizeof *otf) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
|
||
otf = ofpbuf_data(msg);
|
||
len = ntohs(otf->length);
|
||
if (len < sizeof *otf || len % 8 || len > ofpbuf_size(msg)) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
ofpbuf_use_const(&properties, ofpbuf_pull(msg, len), len);
|
||
ofpbuf_pull(&properties, sizeof *otf);
|
||
|
||
tf->table_id = otf->table_id;
|
||
if (tf->table_id == OFPTT_ALL) {
|
||
return OFPERR_OFPTFFC_BAD_TABLE;
|
||
}
|
||
|
||
ovs_strlcpy(tf->name, otf->name, OFP_MAX_TABLE_NAME_LEN);
|
||
tf->metadata_match = otf->metadata_match;
|
||
tf->metadata_write = otf->metadata_write;
|
||
tf->miss_config = ofputil_table_miss_from_config(otf->config, oh->version);
|
||
tf->max_entries = ntohl(otf->max_entries);
|
||
|
||
while (ofpbuf_size(&properties) > 0) {
|
||
struct ofpbuf payload;
|
||
enum ofperr error;
|
||
uint16_t type;
|
||
|
||
error = pull_table_feature_property(&properties, &payload, &type);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
switch ((enum ofp13_table_feature_prop_type) type) {
|
||
case OFPTFPT13_INSTRUCTIONS:
|
||
error = parse_instruction_ids(&payload, loose,
|
||
&tf->nonmiss.instructions);
|
||
break;
|
||
case OFPTFPT13_INSTRUCTIONS_MISS:
|
||
error = parse_instruction_ids(&payload, loose,
|
||
&tf->miss.instructions);
|
||
break;
|
||
|
||
case OFPTFPT13_NEXT_TABLES:
|
||
error = parse_table_features_next_table(&payload,
|
||
tf->nonmiss.next);
|
||
break;
|
||
case OFPTFPT13_NEXT_TABLES_MISS:
|
||
error = parse_table_features_next_table(&payload, tf->miss.next);
|
||
break;
|
||
|
||
case OFPTFPT13_WRITE_ACTIONS:
|
||
error = parse_action_bitmap(&payload, oh->version,
|
||
&tf->nonmiss.write.ofpacts);
|
||
break;
|
||
case OFPTFPT13_WRITE_ACTIONS_MISS:
|
||
error = parse_action_bitmap(&payload, oh->version,
|
||
&tf->miss.write.ofpacts);
|
||
break;
|
||
|
||
case OFPTFPT13_APPLY_ACTIONS:
|
||
error = parse_action_bitmap(&payload, oh->version,
|
||
&tf->nonmiss.apply.ofpacts);
|
||
break;
|
||
case OFPTFPT13_APPLY_ACTIONS_MISS:
|
||
error = parse_action_bitmap(&payload, oh->version,
|
||
&tf->miss.apply.ofpacts);
|
||
break;
|
||
|
||
case OFPTFPT13_MATCH:
|
||
error = parse_oxms(&payload, loose, &tf->match, &tf->mask);
|
||
break;
|
||
case OFPTFPT13_WILDCARDS:
|
||
error = parse_oxms(&payload, loose, &tf->wildcard, NULL);
|
||
break;
|
||
|
||
case OFPTFPT13_WRITE_SETFIELD:
|
||
error = parse_oxms(&payload, loose,
|
||
&tf->nonmiss.write.set_fields, NULL);
|
||
break;
|
||
case OFPTFPT13_WRITE_SETFIELD_MISS:
|
||
error = parse_oxms(&payload, loose,
|
||
&tf->miss.write.set_fields, NULL);
|
||
break;
|
||
case OFPTFPT13_APPLY_SETFIELD:
|
||
error = parse_oxms(&payload, loose,
|
||
&tf->nonmiss.apply.set_fields, NULL);
|
||
break;
|
||
case OFPTFPT13_APPLY_SETFIELD_MISS:
|
||
error = parse_oxms(&payload, loose,
|
||
&tf->miss.apply.set_fields, NULL);
|
||
break;
|
||
|
||
case OFPTFPT13_EXPERIMENTER:
|
||
case OFPTFPT13_EXPERIMENTER_MISS:
|
||
default:
|
||
log_property(loose, "unknown table features property %"PRIu16,
|
||
type);
|
||
error = loose ? 0 : OFPERR_OFPBPC_BAD_TYPE;
|
||
break;
|
||
}
|
||
if (error) {
|
||
return error;
|
||
}
|
||
}
|
||
|
||
/* Fix inconsistencies:
|
||
*
|
||
* - Turn on 'match' bits that are set in 'mask', because maskable
|
||
* fields are matchable.
|
||
*
|
||
* - Turn on 'wildcard' bits that are set in 'mask', because a field
|
||
* that is arbitrarily maskable can be wildcarded entirely.
|
||
*
|
||
* - Turn off 'wildcard' bits that are not in 'match', because a field
|
||
* must be matchable for it to be meaningfully wildcarded. */
|
||
bitmap_or(tf->match.bm, tf->mask.bm, MFF_N_IDS);
|
||
bitmap_or(tf->wildcard.bm, tf->mask.bm, MFF_N_IDS);
|
||
bitmap_and(tf->wildcard.bm, tf->match.bm, MFF_N_IDS);
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Encodes and returns a request to obtain the table features of a switch.
|
||
* The message is encoded for OpenFlow version 'ofp_version'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_table_features_request(enum ofp_version ofp_version)
|
||
{
|
||
struct ofpbuf *request = NULL;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
ovs_fatal(0, "dump-table-features needs OpenFlow 1.3 or later "
|
||
"(\'-O OpenFlow13\')");
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST13_TABLE_FEATURES_REQUEST,
|
||
ofp_version, 0);
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
static void
|
||
put_fields_property(struct ofpbuf *reply,
|
||
const struct mf_bitmap *fields,
|
||
const struct mf_bitmap *masks,
|
||
enum ofp13_table_feature_prop_type property,
|
||
enum ofp_version version)
|
||
{
|
||
size_t start_ofs;
|
||
int field;
|
||
|
||
start_ofs = start_property(reply, property);
|
||
BITMAP_FOR_EACH_1 (field, MFF_N_IDS, fields->bm) {
|
||
uint32_t h_oxm = mf_oxm_header(field, version);
|
||
ovs_be32 n_oxm;
|
||
|
||
if (masks && bitmap_is_set(masks->bm, field)) {
|
||
h_oxm = NXM_MAKE_WILD_HEADER(h_oxm);
|
||
}
|
||
|
||
n_oxm = htonl(h_oxm);
|
||
ofpbuf_put(reply, &n_oxm, sizeof n_oxm);
|
||
}
|
||
end_property(reply, start_ofs);
|
||
}
|
||
|
||
static void
|
||
put_table_action_features(struct ofpbuf *reply,
|
||
const struct ofputil_table_action_features *taf,
|
||
enum ofp13_table_feature_prop_type actions_type,
|
||
enum ofp13_table_feature_prop_type set_fields_type,
|
||
int miss_offset, enum ofp_version version)
|
||
{
|
||
size_t start_ofs;
|
||
|
||
start_ofs = start_property(reply, actions_type + miss_offset);
|
||
put_bitmap_properties(reply,
|
||
ntohl(ofpact_bitmap_to_openflow(taf->ofpacts,
|
||
version)));
|
||
end_property(reply, start_ofs);
|
||
|
||
put_fields_property(reply, &taf->set_fields, NULL,
|
||
set_fields_type + miss_offset, version);
|
||
}
|
||
|
||
static void
|
||
put_table_instruction_features(
|
||
struct ofpbuf *reply, const struct ofputil_table_instruction_features *tif,
|
||
int miss_offset, enum ofp_version version)
|
||
{
|
||
size_t start_ofs;
|
||
uint8_t table_id;
|
||
|
||
start_ofs = start_property(reply, OFPTFPT13_INSTRUCTIONS + miss_offset);
|
||
put_bitmap_properties(reply,
|
||
ntohl(ovsinst_bitmap_to_openflow(tif->instructions,
|
||
version)));
|
||
end_property(reply, start_ofs);
|
||
|
||
start_ofs = start_property(reply, OFPTFPT13_NEXT_TABLES + miss_offset);
|
||
BITMAP_FOR_EACH_1 (table_id, 255, tif->next) {
|
||
ofpbuf_put(reply, &table_id, 1);
|
||
}
|
||
end_property(reply, start_ofs);
|
||
|
||
put_table_action_features(reply, &tif->write,
|
||
OFPTFPT13_WRITE_ACTIONS,
|
||
OFPTFPT13_WRITE_SETFIELD, miss_offset, version);
|
||
put_table_action_features(reply, &tif->apply,
|
||
OFPTFPT13_APPLY_ACTIONS,
|
||
OFPTFPT13_APPLY_SETFIELD, miss_offset, version);
|
||
}
|
||
|
||
void
|
||
ofputil_append_table_features_reply(const struct ofputil_table_features *tf,
|
||
struct list *replies)
|
||
{
|
||
struct ofpbuf *reply = ofpbuf_from_list(list_back(replies));
|
||
enum ofp_version version = ofpmp_version(replies);
|
||
size_t start_ofs = ofpbuf_size(reply);
|
||
struct ofp13_table_features *otf;
|
||
|
||
otf = ofpbuf_put_zeros(reply, sizeof *otf);
|
||
otf->table_id = tf->table_id;
|
||
ovs_strlcpy(otf->name, tf->name, sizeof otf->name);
|
||
otf->metadata_match = tf->metadata_match;
|
||
otf->metadata_write = tf->metadata_write;
|
||
otf->config = ofputil_table_miss_to_config(tf->miss_config, version);
|
||
otf->max_entries = htonl(tf->max_entries);
|
||
|
||
put_table_instruction_features(reply, &tf->nonmiss, 0, version);
|
||
put_table_instruction_features(reply, &tf->miss, 1, version);
|
||
|
||
put_fields_property(reply, &tf->match, &tf->mask,
|
||
OFPTFPT13_MATCH, version);
|
||
put_fields_property(reply, &tf->wildcard, NULL,
|
||
OFPTFPT13_WILDCARDS, version);
|
||
|
||
otf = ofpbuf_at_assert(reply, start_ofs, sizeof *otf);
|
||
otf->length = htons(ofpbuf_size(reply) - start_ofs);
|
||
ofpmp_postappend(replies, start_ofs);
|
||
}
|
||
|
||
/* ofputil_table_mod */
|
||
|
||
/* Given 'config', taken from an OpenFlow 'version' message that specifies
|
||
* table configuration (a table mod, table stats, or table features message),
|
||
* returns the table miss configuration that it specifies. */
|
||
static enum ofputil_table_miss
|
||
ofputil_table_miss_from_config(ovs_be32 config_, enum ofp_version version)
|
||
{
|
||
uint32_t config = ntohl(config_);
|
||
|
||
if (version < OFP13_VERSION) {
|
||
switch (config & OFPTC11_TABLE_MISS_MASK) {
|
||
case OFPTC11_TABLE_MISS_CONTROLLER:
|
||
return OFPUTIL_TABLE_MISS_CONTROLLER;
|
||
|
||
case OFPTC11_TABLE_MISS_CONTINUE:
|
||
return OFPUTIL_TABLE_MISS_CONTINUE;
|
||
|
||
case OFPTC11_TABLE_MISS_DROP:
|
||
return OFPUTIL_TABLE_MISS_DROP;
|
||
|
||
default:
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "bad table miss config %d", config);
|
||
return OFPUTIL_TABLE_MISS_CONTROLLER;
|
||
}
|
||
} else {
|
||
return OFPUTIL_TABLE_MISS_DEFAULT;
|
||
}
|
||
}
|
||
|
||
/* Given a table miss configuration, returns the corresponding OpenFlow table
|
||
* configuration for use in an OpenFlow message of the given 'version'. */
|
||
ovs_be32
|
||
ofputil_table_miss_to_config(enum ofputil_table_miss miss,
|
||
enum ofp_version version)
|
||
{
|
||
if (version < OFP13_VERSION) {
|
||
switch (miss) {
|
||
case OFPUTIL_TABLE_MISS_CONTROLLER:
|
||
case OFPUTIL_TABLE_MISS_DEFAULT:
|
||
return htonl(OFPTC11_TABLE_MISS_CONTROLLER);
|
||
|
||
case OFPUTIL_TABLE_MISS_CONTINUE:
|
||
return htonl(OFPTC11_TABLE_MISS_CONTINUE);
|
||
|
||
case OFPUTIL_TABLE_MISS_DROP:
|
||
return htonl(OFPTC11_TABLE_MISS_DROP);
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
} else {
|
||
return htonl(0);
|
||
}
|
||
}
|
||
|
||
/* Decodes the OpenFlow "table mod" message in '*oh' into an abstract form in
|
||
* '*pm'. Returns 0 if successful, otherwise an OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_table_mod(const struct ofp_header *oh,
|
||
struct ofputil_table_mod *pm)
|
||
{
|
||
enum ofpraw raw;
|
||
struct ofpbuf b;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
if (raw == OFPRAW_OFPT11_TABLE_MOD) {
|
||
const struct ofp11_table_mod *otm = ofpbuf_data(&b);
|
||
|
||
pm->table_id = otm->table_id;
|
||
pm->miss_config = ofputil_table_miss_from_config(otm->config,
|
||
oh->version);
|
||
} else if (raw == OFPRAW_OFPT14_TABLE_MOD) {
|
||
const struct ofp14_table_mod *otm = ofpbuf_pull(&b, sizeof *otm);
|
||
|
||
pm->table_id = otm->table_id;
|
||
pm->miss_config = ofputil_table_miss_from_config(otm->config,
|
||
oh->version);
|
||
/* We do not understand any properties yet, so we do not bother
|
||
* parsing them. */
|
||
} else {
|
||
return OFPERR_OFPBRC_BAD_TYPE;
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Converts the abstract form of a "table mod" message in '*pm' into an OpenFlow
|
||
* message suitable for 'protocol', and returns that encoded form in a buffer
|
||
* owned by the caller. */
|
||
struct ofpbuf *
|
||
ofputil_encode_table_mod(const struct ofputil_table_mod *pm,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
enum ofp_version ofp_version = ofputil_protocol_to_ofp_version(protocol);
|
||
struct ofpbuf *b;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
ovs_fatal(0, "table mod needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
break;
|
||
}
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION: {
|
||
struct ofp11_table_mod *otm;
|
||
|
||
b = ofpraw_alloc(OFPRAW_OFPT11_TABLE_MOD, ofp_version, 0);
|
||
otm = ofpbuf_put_zeros(b, sizeof *otm);
|
||
otm->table_id = pm->table_id;
|
||
otm->config = ofputil_table_miss_to_config(pm->miss_config,
|
||
ofp_version);
|
||
break;
|
||
}
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp14_table_mod *otm;
|
||
|
||
b = ofpraw_alloc(OFPRAW_OFPT14_TABLE_MOD, ofp_version, 0);
|
||
otm = ofpbuf_put_zeros(b, sizeof *otm);
|
||
otm->table_id = pm->table_id;
|
||
otm->config = ofputil_table_miss_to_config(pm->miss_config,
|
||
ofp_version);
|
||
break;
|
||
}
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return b;
|
||
}
|
||
|
||
/* ofputil_role_request */
|
||
|
||
/* Decodes the OpenFlow "role request" or "role reply" message in '*oh' into
|
||
* an abstract form in '*rr'. Returns 0 if successful, otherwise an
|
||
* OFPERR_* value. */
|
||
enum ofperr
|
||
ofputil_decode_role_message(const struct ofp_header *oh,
|
||
struct ofputil_role_request *rr)
|
||
{
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
|
||
if (raw == OFPRAW_OFPT12_ROLE_REQUEST ||
|
||
raw == OFPRAW_OFPT12_ROLE_REPLY) {
|
||
const struct ofp12_role_request *orr = ofpbuf_l3(&b);
|
||
|
||
if (orr->role != htonl(OFPCR12_ROLE_NOCHANGE) &&
|
||
orr->role != htonl(OFPCR12_ROLE_EQUAL) &&
|
||
orr->role != htonl(OFPCR12_ROLE_MASTER) &&
|
||
orr->role != htonl(OFPCR12_ROLE_SLAVE)) {
|
||
return OFPERR_OFPRRFC_BAD_ROLE;
|
||
}
|
||
|
||
rr->role = ntohl(orr->role);
|
||
if (raw == OFPRAW_OFPT12_ROLE_REQUEST
|
||
? orr->role == htonl(OFPCR12_ROLE_NOCHANGE)
|
||
: orr->generation_id == OVS_BE64_MAX) {
|
||
rr->have_generation_id = false;
|
||
rr->generation_id = 0;
|
||
} else {
|
||
rr->have_generation_id = true;
|
||
rr->generation_id = ntohll(orr->generation_id);
|
||
}
|
||
} else if (raw == OFPRAW_NXT_ROLE_REQUEST ||
|
||
raw == OFPRAW_NXT_ROLE_REPLY) {
|
||
const struct nx_role_request *nrr = ofpbuf_l3(&b);
|
||
|
||
BUILD_ASSERT(NX_ROLE_OTHER + 1 == OFPCR12_ROLE_EQUAL);
|
||
BUILD_ASSERT(NX_ROLE_MASTER + 1 == OFPCR12_ROLE_MASTER);
|
||
BUILD_ASSERT(NX_ROLE_SLAVE + 1 == OFPCR12_ROLE_SLAVE);
|
||
|
||
if (nrr->role != htonl(NX_ROLE_OTHER) &&
|
||
nrr->role != htonl(NX_ROLE_MASTER) &&
|
||
nrr->role != htonl(NX_ROLE_SLAVE)) {
|
||
return OFPERR_OFPRRFC_BAD_ROLE;
|
||
}
|
||
|
||
rr->role = ntohl(nrr->role) + 1;
|
||
rr->have_generation_id = false;
|
||
rr->generation_id = 0;
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Returns an encoded form of a role reply suitable for the "request" in a
|
||
* buffer owned by the caller. */
|
||
struct ofpbuf *
|
||
ofputil_encode_role_reply(const struct ofp_header *request,
|
||
const struct ofputil_role_request *rr)
|
||
{
|
||
struct ofpbuf *buf;
|
||
enum ofpraw raw;
|
||
|
||
raw = ofpraw_decode_assert(request);
|
||
if (raw == OFPRAW_OFPT12_ROLE_REQUEST) {
|
||
struct ofp12_role_request *orr;
|
||
|
||
buf = ofpraw_alloc_reply(OFPRAW_OFPT12_ROLE_REPLY, request, 0);
|
||
orr = ofpbuf_put_zeros(buf, sizeof *orr);
|
||
|
||
orr->role = htonl(rr->role);
|
||
orr->generation_id = htonll(rr->have_generation_id
|
||
? rr->generation_id
|
||
: UINT64_MAX);
|
||
} else if (raw == OFPRAW_NXT_ROLE_REQUEST) {
|
||
struct nx_role_request *nrr;
|
||
|
||
BUILD_ASSERT(NX_ROLE_OTHER == OFPCR12_ROLE_EQUAL - 1);
|
||
BUILD_ASSERT(NX_ROLE_MASTER == OFPCR12_ROLE_MASTER - 1);
|
||
BUILD_ASSERT(NX_ROLE_SLAVE == OFPCR12_ROLE_SLAVE - 1);
|
||
|
||
buf = ofpraw_alloc_reply(OFPRAW_NXT_ROLE_REPLY, request, 0);
|
||
nrr = ofpbuf_put_zeros(buf, sizeof *nrr);
|
||
nrr->role = htonl(rr->role - 1);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return buf;
|
||
}
|
||
|
||
/* Encodes "role status" message 'status' for sending in the given
|
||
* 'protocol'. Returns the role status message, if 'protocol' supports them,
|
||
* otherwise a null pointer. */
|
||
struct ofpbuf *
|
||
ofputil_encode_role_status(const struct ofputil_role_status *status,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
enum ofp_version version;
|
||
|
||
version = ofputil_protocol_to_ofp_version(protocol);
|
||
if (version >= OFP14_VERSION) {
|
||
struct ofp14_role_status *rstatus;
|
||
struct ofpbuf *buf;
|
||
|
||
buf = ofpraw_alloc_xid(OFPRAW_OFPT14_ROLE_STATUS, version, htonl(0),
|
||
0);
|
||
rstatus = ofpbuf_put_zeros(buf, sizeof *rstatus);
|
||
rstatus->role = htonl(status->role);
|
||
rstatus->reason = status->reason;
|
||
rstatus->generation_id = htonll(status->generation_id);
|
||
|
||
return buf;
|
||
} else {
|
||
return NULL;
|
||
}
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_role_status(const struct ofp_header *oh,
|
||
struct ofputil_role_status *rs)
|
||
{
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
const struct ofp14_role_status *r;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
ovs_assert(raw == OFPRAW_OFPT14_ROLE_STATUS);
|
||
|
||
r = ofpbuf_l3(&b);
|
||
if (r->role != htonl(OFPCR12_ROLE_NOCHANGE) &&
|
||
r->role != htonl(OFPCR12_ROLE_EQUAL) &&
|
||
r->role != htonl(OFPCR12_ROLE_MASTER) &&
|
||
r->role != htonl(OFPCR12_ROLE_SLAVE)) {
|
||
return OFPERR_OFPRRFC_BAD_ROLE;
|
||
}
|
||
|
||
rs->role = ntohl(r->role);
|
||
rs->generation_id = ntohll(r->generation_id);
|
||
rs->reason = r->reason;
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Table stats. */
|
||
|
||
/* OpenFlow 1.0 and 1.1 don't distinguish between a field that cannot be
|
||
* matched and a field that must be wildcarded. This function returns a bitmap
|
||
* that contains both kinds of fields. */
|
||
static struct mf_bitmap
|
||
wild_or_nonmatchable_fields(const struct ofputil_table_features *features)
|
||
{
|
||
struct mf_bitmap wc = features->match;
|
||
bitmap_not(wc.bm, MFF_N_IDS);
|
||
bitmap_or(wc.bm, features->wildcard.bm, MFF_N_IDS);
|
||
return wc;
|
||
}
|
||
|
||
struct ofp10_wc_map {
|
||
enum ofp10_flow_wildcards wc10;
|
||
enum mf_field_id mf;
|
||
};
|
||
|
||
static const struct ofp10_wc_map ofp10_wc_map[] = {
|
||
{ OFPFW10_IN_PORT, MFF_IN_PORT },
|
||
{ OFPFW10_DL_VLAN, MFF_VLAN_VID },
|
||
{ OFPFW10_DL_SRC, MFF_ETH_SRC },
|
||
{ OFPFW10_DL_DST, MFF_ETH_DST},
|
||
{ OFPFW10_DL_TYPE, MFF_ETH_TYPE },
|
||
{ OFPFW10_NW_PROTO, MFF_IP_PROTO },
|
||
{ OFPFW10_TP_SRC, MFF_TCP_SRC },
|
||
{ OFPFW10_TP_DST, MFF_TCP_DST },
|
||
{ OFPFW10_NW_SRC_MASK, MFF_IPV4_SRC },
|
||
{ OFPFW10_NW_DST_MASK, MFF_IPV4_DST },
|
||
{ OFPFW10_DL_VLAN_PCP, MFF_VLAN_PCP },
|
||
{ OFPFW10_NW_TOS, MFF_IP_DSCP },
|
||
};
|
||
|
||
static ovs_be32
|
||
mf_bitmap_to_of10(const struct mf_bitmap *fields)
|
||
{
|
||
const struct ofp10_wc_map *p;
|
||
uint32_t wc10 = 0;
|
||
|
||
for (p = ofp10_wc_map; p < &ofp10_wc_map[ARRAY_SIZE(ofp10_wc_map)]; p++) {
|
||
if (bitmap_is_set(fields->bm, p->mf)) {
|
||
wc10 |= p->wc10;
|
||
}
|
||
}
|
||
return htonl(wc10);
|
||
}
|
||
|
||
static struct mf_bitmap
|
||
mf_bitmap_from_of10(ovs_be32 wc10_)
|
||
{
|
||
struct mf_bitmap fields = MF_BITMAP_INITIALIZER;
|
||
const struct ofp10_wc_map *p;
|
||
uint32_t wc10 = ntohl(wc10_);
|
||
|
||
for (p = ofp10_wc_map; p < &ofp10_wc_map[ARRAY_SIZE(ofp10_wc_map)]; p++) {
|
||
if (wc10 & p->wc10) {
|
||
bitmap_set1(fields.bm, p->mf);
|
||
}
|
||
}
|
||
return fields;
|
||
}
|
||
|
||
static void
|
||
ofputil_put_ofp10_table_stats(const struct ofputil_table_stats *stats,
|
||
const struct ofputil_table_features *features,
|
||
struct ofpbuf *buf)
|
||
{
|
||
struct mf_bitmap wc = wild_or_nonmatchable_fields(features);
|
||
struct ofp10_table_stats *out;
|
||
|
||
out = ofpbuf_put_zeros(buf, sizeof *out);
|
||
out->table_id = features->table_id;
|
||
ovs_strlcpy(out->name, features->name, sizeof out->name);
|
||
out->wildcards = mf_bitmap_to_of10(&wc);
|
||
out->max_entries = htonl(features->max_entries);
|
||
out->active_count = htonl(stats->active_count);
|
||
put_32aligned_be64(&out->lookup_count, htonll(stats->lookup_count));
|
||
put_32aligned_be64(&out->matched_count, htonll(stats->matched_count));
|
||
}
|
||
|
||
struct ofp11_wc_map {
|
||
enum ofp11_flow_match_fields wc11;
|
||
enum mf_field_id mf;
|
||
};
|
||
|
||
static const struct ofp11_wc_map ofp11_wc_map[] = {
|
||
{ OFPFMF11_IN_PORT, MFF_IN_PORT },
|
||
{ OFPFMF11_DL_VLAN, MFF_VLAN_VID },
|
||
{ OFPFMF11_DL_VLAN_PCP, MFF_VLAN_PCP },
|
||
{ OFPFMF11_DL_TYPE, MFF_ETH_TYPE },
|
||
{ OFPFMF11_NW_TOS, MFF_IP_DSCP },
|
||
{ OFPFMF11_NW_PROTO, MFF_IP_PROTO },
|
||
{ OFPFMF11_TP_SRC, MFF_TCP_SRC },
|
||
{ OFPFMF11_TP_DST, MFF_TCP_DST },
|
||
{ OFPFMF11_MPLS_LABEL, MFF_MPLS_LABEL },
|
||
{ OFPFMF11_MPLS_TC, MFF_MPLS_TC },
|
||
/* I don't know what OFPFMF11_TYPE means. */
|
||
{ OFPFMF11_DL_SRC, MFF_ETH_SRC },
|
||
{ OFPFMF11_DL_DST, MFF_ETH_DST },
|
||
{ OFPFMF11_NW_SRC, MFF_IPV4_SRC },
|
||
{ OFPFMF11_NW_DST, MFF_IPV4_DST },
|
||
{ OFPFMF11_METADATA, MFF_METADATA },
|
||
};
|
||
|
||
static ovs_be32
|
||
mf_bitmap_to_of11(const struct mf_bitmap *fields)
|
||
{
|
||
const struct ofp11_wc_map *p;
|
||
uint32_t wc11 = 0;
|
||
|
||
for (p = ofp11_wc_map; p < &ofp11_wc_map[ARRAY_SIZE(ofp11_wc_map)]; p++) {
|
||
if (bitmap_is_set(fields->bm, p->mf)) {
|
||
wc11 |= p->wc11;
|
||
}
|
||
}
|
||
return htonl(wc11);
|
||
}
|
||
|
||
static struct mf_bitmap
|
||
mf_bitmap_from_of11(ovs_be32 wc11_)
|
||
{
|
||
struct mf_bitmap fields = MF_BITMAP_INITIALIZER;
|
||
const struct ofp11_wc_map *p;
|
||
uint32_t wc11 = ntohl(wc11_);
|
||
|
||
for (p = ofp11_wc_map; p < &ofp11_wc_map[ARRAY_SIZE(ofp11_wc_map)]; p++) {
|
||
if (wc11 & p->wc11) {
|
||
bitmap_set1(fields.bm, p->mf);
|
||
}
|
||
}
|
||
return fields;
|
||
}
|
||
|
||
static void
|
||
ofputil_put_ofp11_table_stats(const struct ofputil_table_stats *stats,
|
||
const struct ofputil_table_features *features,
|
||
struct ofpbuf *buf)
|
||
{
|
||
struct mf_bitmap wc = wild_or_nonmatchable_fields(features);
|
||
struct ofp11_table_stats *out;
|
||
|
||
out = ofpbuf_put_zeros(buf, sizeof *out);
|
||
out->table_id = features->table_id;
|
||
ovs_strlcpy(out->name, features->name, sizeof out->name);
|
||
out->wildcards = mf_bitmap_to_of11(&wc);
|
||
out->match = mf_bitmap_to_of11(&features->match);
|
||
out->instructions = ovsinst_bitmap_to_openflow(
|
||
features->nonmiss.instructions, OFP11_VERSION);
|
||
out->write_actions = ofpact_bitmap_to_openflow(
|
||
features->nonmiss.write.ofpacts, OFP11_VERSION);
|
||
out->apply_actions = ofpact_bitmap_to_openflow(
|
||
features->nonmiss.apply.ofpacts, OFP11_VERSION);
|
||
out->config = htonl(features->miss_config);
|
||
out->max_entries = htonl(features->max_entries);
|
||
out->active_count = htonl(stats->active_count);
|
||
out->lookup_count = htonll(stats->lookup_count);
|
||
out->matched_count = htonll(stats->matched_count);
|
||
}
|
||
|
||
static ovs_be64
|
||
mf_bitmap_to_oxm_bitmap(const struct mf_bitmap *fields,
|
||
enum ofp_version version)
|
||
{
|
||
uint64_t oxm_bitmap = 0;
|
||
int i;
|
||
|
||
BITMAP_FOR_EACH_1 (i, MFF_N_IDS, fields->bm) {
|
||
uint32_t oxm = mf_oxm_header(i, version);
|
||
uint32_t vendor = NXM_VENDOR(oxm);
|
||
int field = NXM_FIELD(oxm);
|
||
|
||
if (vendor == OFPXMC12_OPENFLOW_BASIC && field < 64) {
|
||
oxm_bitmap |= UINT64_C(1) << field;
|
||
}
|
||
}
|
||
return htonll(oxm_bitmap);
|
||
}
|
||
|
||
static struct mf_bitmap
|
||
mf_bitmap_from_oxm_bitmap(ovs_be64 oxm_bitmap, enum ofp_version version)
|
||
{
|
||
struct mf_bitmap fields = MF_BITMAP_INITIALIZER;
|
||
|
||
for (enum mf_field_id id = 0; id < MFF_N_IDS; id++) {
|
||
const struct mf_field *f = mf_from_id(id);
|
||
uint32_t oxm = f->oxm_header;
|
||
uint32_t vendor = NXM_VENDOR(oxm);
|
||
int field = NXM_FIELD(oxm);
|
||
|
||
if (version >= f->oxm_version
|
||
&& vendor == OFPXMC12_OPENFLOW_BASIC
|
||
&& field < 64
|
||
&& oxm_bitmap & htonll(UINT64_C(1) << field)) {
|
||
bitmap_set1(fields.bm, id);
|
||
}
|
||
}
|
||
return fields;
|
||
}
|
||
|
||
static void
|
||
ofputil_put_ofp12_table_stats(const struct ofputil_table_stats *stats,
|
||
const struct ofputil_table_features *features,
|
||
struct ofpbuf *buf)
|
||
{
|
||
struct ofp12_table_stats *out;
|
||
|
||
out = ofpbuf_put_zeros(buf, sizeof *out);
|
||
out->table_id = features->table_id;
|
||
ovs_strlcpy(out->name, features->name, sizeof out->name);
|
||
out->match = mf_bitmap_to_oxm_bitmap(&features->match, OFP12_VERSION);
|
||
out->wildcards = mf_bitmap_to_oxm_bitmap(&features->wildcard,
|
||
OFP12_VERSION);
|
||
out->write_actions = ofpact_bitmap_to_openflow(
|
||
features->nonmiss.write.ofpacts, OFP12_VERSION);
|
||
out->apply_actions = ofpact_bitmap_to_openflow(
|
||
features->nonmiss.apply.ofpacts, OFP12_VERSION);
|
||
out->write_setfields = mf_bitmap_to_oxm_bitmap(
|
||
&features->nonmiss.write.set_fields, OFP12_VERSION);
|
||
out->apply_setfields = mf_bitmap_to_oxm_bitmap(
|
||
&features->nonmiss.apply.set_fields, OFP12_VERSION);
|
||
out->metadata_match = features->metadata_match;
|
||
out->metadata_write = features->metadata_write;
|
||
out->instructions = ovsinst_bitmap_to_openflow(
|
||
features->nonmiss.instructions, OFP12_VERSION);
|
||
out->config = ofputil_table_miss_to_config(features->miss_config,
|
||
OFP12_VERSION);
|
||
out->max_entries = htonl(features->max_entries);
|
||
out->active_count = htonl(stats->active_count);
|
||
out->lookup_count = htonll(stats->lookup_count);
|
||
out->matched_count = htonll(stats->matched_count);
|
||
}
|
||
|
||
static void
|
||
ofputil_put_ofp13_table_stats(const struct ofputil_table_stats *stats,
|
||
struct ofpbuf *buf)
|
||
{
|
||
struct ofp13_table_stats *out;
|
||
|
||
out = ofpbuf_put_zeros(buf, sizeof *out);
|
||
out->table_id = stats->table_id;
|
||
out->active_count = htonl(stats->active_count);
|
||
out->lookup_count = htonll(stats->lookup_count);
|
||
out->matched_count = htonll(stats->matched_count);
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_table_stats_reply(const struct ofp_header *request)
|
||
{
|
||
return ofpraw_alloc_stats_reply(request, 0);
|
||
}
|
||
|
||
void
|
||
ofputil_append_table_stats_reply(struct ofpbuf *reply,
|
||
const struct ofputil_table_stats *stats,
|
||
const struct ofputil_table_features *features)
|
||
{
|
||
struct ofp_header *oh = ofpbuf_l2(reply);
|
||
|
||
ovs_assert(stats->table_id == features->table_id);
|
||
|
||
switch ((enum ofp_version) oh->version) {
|
||
case OFP10_VERSION:
|
||
ofputil_put_ofp10_table_stats(stats, features, reply);
|
||
break;
|
||
|
||
case OFP11_VERSION:
|
||
ofputil_put_ofp11_table_stats(stats, features, reply);
|
||
break;
|
||
|
||
case OFP12_VERSION:
|
||
ofputil_put_ofp12_table_stats(stats, features, reply);
|
||
break;
|
||
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
ofputil_put_ofp13_table_stats(stats, reply);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
static int
|
||
ofputil_decode_ofp10_table_stats(struct ofpbuf *msg,
|
||
struct ofputil_table_stats *stats,
|
||
struct ofputil_table_features *features)
|
||
{
|
||
struct ofp10_table_stats *ots;
|
||
|
||
ots = ofpbuf_try_pull(msg, sizeof *ots);
|
||
if (!ots) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
features->table_id = ots->table_id;
|
||
ovs_strlcpy(features->name, ots->name, sizeof features->name);
|
||
features->max_entries = ntohl(ots->max_entries);
|
||
features->match = features->wildcard = mf_bitmap_from_of10(ots->wildcards);
|
||
|
||
stats->table_id = ots->table_id;
|
||
stats->active_count = ntohl(ots->active_count);
|
||
stats->lookup_count = ntohll(get_32aligned_be64(&ots->lookup_count));
|
||
stats->matched_count = ntohll(get_32aligned_be64(&ots->matched_count));
|
||
|
||
return 0;
|
||
}
|
||
|
||
static int
|
||
ofputil_decode_ofp11_table_stats(struct ofpbuf *msg,
|
||
struct ofputil_table_stats *stats,
|
||
struct ofputil_table_features *features)
|
||
{
|
||
struct ofp11_table_stats *ots;
|
||
|
||
ots = ofpbuf_try_pull(msg, sizeof *ots);
|
||
if (!ots) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
features->table_id = ots->table_id;
|
||
ovs_strlcpy(features->name, ots->name, sizeof features->name);
|
||
features->max_entries = ntohl(ots->max_entries);
|
||
features->nonmiss.instructions = ovsinst_bitmap_from_openflow(
|
||
ots->instructions, OFP11_VERSION);
|
||
features->nonmiss.write.ofpacts = ofpact_bitmap_from_openflow(
|
||
ots->write_actions, OFP11_VERSION);
|
||
features->nonmiss.apply.ofpacts = ofpact_bitmap_from_openflow(
|
||
ots->write_actions, OFP11_VERSION);
|
||
features->miss = features->nonmiss;
|
||
features->miss_config = ofputil_table_miss_from_config(ots->config,
|
||
OFP11_VERSION);
|
||
features->match = mf_bitmap_from_of11(ots->match);
|
||
features->wildcard = mf_bitmap_from_of11(ots->wildcards);
|
||
bitmap_or(features->match.bm, features->wildcard.bm, MFF_N_IDS);
|
||
|
||
stats->table_id = ots->table_id;
|
||
stats->active_count = ntohl(ots->active_count);
|
||
stats->lookup_count = ntohll(ots->lookup_count);
|
||
stats->matched_count = ntohll(ots->matched_count);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static int
|
||
ofputil_decode_ofp12_table_stats(struct ofpbuf *msg,
|
||
struct ofputil_table_stats *stats,
|
||
struct ofputil_table_features *features)
|
||
{
|
||
struct ofp12_table_stats *ots;
|
||
|
||
ots = ofpbuf_try_pull(msg, sizeof *ots);
|
||
if (!ots) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
features->table_id = ots->table_id;
|
||
ovs_strlcpy(features->name, ots->name, sizeof features->name);
|
||
features->metadata_match = ots->metadata_match;
|
||
features->metadata_write = ots->metadata_write;
|
||
features->miss_config = ofputil_table_miss_from_config(ots->config,
|
||
OFP12_VERSION);
|
||
features->max_entries = ntohl(ots->max_entries);
|
||
|
||
features->nonmiss.instructions = ovsinst_bitmap_from_openflow(
|
||
ots->instructions, OFP12_VERSION);
|
||
features->nonmiss.write.ofpacts = ofpact_bitmap_from_openflow(
|
||
ots->write_actions, OFP12_VERSION);
|
||
features->nonmiss.apply.ofpacts = ofpact_bitmap_from_openflow(
|
||
ots->apply_actions, OFP12_VERSION);
|
||
features->nonmiss.write.set_fields = mf_bitmap_from_oxm_bitmap(
|
||
ots->write_setfields, OFP12_VERSION);
|
||
features->nonmiss.apply.set_fields = mf_bitmap_from_oxm_bitmap(
|
||
ots->apply_setfields, OFP12_VERSION);
|
||
features->miss = features->nonmiss;
|
||
|
||
features->match = mf_bitmap_from_oxm_bitmap(ots->match, OFP12_VERSION);
|
||
features->wildcard = mf_bitmap_from_oxm_bitmap(ots->wildcards,
|
||
OFP12_VERSION);
|
||
bitmap_or(features->match.bm, features->wildcard.bm, MFF_N_IDS);
|
||
|
||
stats->table_id = ots->table_id;
|
||
stats->active_count = ntohl(ots->active_count);
|
||
stats->lookup_count = ntohll(ots->lookup_count);
|
||
stats->matched_count = ntohll(ots->matched_count);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static int
|
||
ofputil_decode_ofp13_table_stats(struct ofpbuf *msg,
|
||
struct ofputil_table_stats *stats,
|
||
struct ofputil_table_features *features)
|
||
{
|
||
struct ofp13_table_stats *ots;
|
||
|
||
ots = ofpbuf_try_pull(msg, sizeof *ots);
|
||
if (!ots) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
features->table_id = ots->table_id;
|
||
|
||
stats->table_id = ots->table_id;
|
||
stats->active_count = ntohl(ots->active_count);
|
||
stats->lookup_count = ntohll(ots->lookup_count);
|
||
stats->matched_count = ntohll(ots->matched_count);
|
||
|
||
return 0;
|
||
}
|
||
|
||
int
|
||
ofputil_decode_table_stats_reply(struct ofpbuf *msg,
|
||
struct ofputil_table_stats *stats,
|
||
struct ofputil_table_features *features)
|
||
{
|
||
const struct ofp_header *oh;
|
||
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
oh = msg->frame;
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
memset(stats, 0, sizeof *stats);
|
||
memset(features, 0, sizeof *features);
|
||
|
||
switch ((enum ofp_version) oh->version) {
|
||
case OFP10_VERSION:
|
||
return ofputil_decode_ofp10_table_stats(msg, stats, features);
|
||
|
||
case OFP11_VERSION:
|
||
return ofputil_decode_ofp11_table_stats(msg, stats, features);
|
||
|
||
case OFP12_VERSION:
|
||
return ofputil_decode_ofp12_table_stats(msg, stats, features);
|
||
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
return ofputil_decode_ofp13_table_stats(msg, stats, features);
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* ofputil_flow_monitor_request */
|
||
|
||
/* Converts an NXST_FLOW_MONITOR request in 'msg' into an abstract
|
||
* ofputil_flow_monitor_request in 'rq'.
|
||
*
|
||
* Multiple NXST_FLOW_MONITOR requests can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the requests. The caller must initially leave 'msg''s layer
|
||
* pointers null and not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no requests were left in this 'msg',
|
||
* otherwise an OFPERR_* value. */
|
||
int
|
||
ofputil_decode_flow_monitor_request(struct ofputil_flow_monitor_request *rq,
|
||
struct ofpbuf *msg)
|
||
{
|
||
struct nx_flow_monitor_request *nfmr;
|
||
uint16_t flags;
|
||
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
nfmr = ofpbuf_try_pull(msg, sizeof *nfmr);
|
||
if (!nfmr) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "NXST_FLOW_MONITOR request has %"PRIu32" "
|
||
"leftover bytes at end", ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
flags = ntohs(nfmr->flags);
|
||
if (!(flags & (NXFMF_ADD | NXFMF_DELETE | NXFMF_MODIFY))
|
||
|| flags & ~(NXFMF_INITIAL | NXFMF_ADD | NXFMF_DELETE
|
||
| NXFMF_MODIFY | NXFMF_ACTIONS | NXFMF_OWN)) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "NXST_FLOW_MONITOR has bad flags %#"PRIx16,
|
||
flags);
|
||
return OFPERR_OFPMOFC_BAD_FLAGS;
|
||
}
|
||
|
||
if (!is_all_zeros(nfmr->zeros, sizeof nfmr->zeros)) {
|
||
return OFPERR_NXBRC_MUST_BE_ZERO;
|
||
}
|
||
|
||
rq->id = ntohl(nfmr->id);
|
||
rq->flags = flags;
|
||
rq->out_port = u16_to_ofp(ntohs(nfmr->out_port));
|
||
rq->table_id = nfmr->table_id;
|
||
|
||
return nx_pull_match(msg, ntohs(nfmr->match_len), &rq->match, NULL, NULL);
|
||
}
|
||
|
||
void
|
||
ofputil_append_flow_monitor_request(
|
||
const struct ofputil_flow_monitor_request *rq, struct ofpbuf *msg)
|
||
{
|
||
struct nx_flow_monitor_request *nfmr;
|
||
size_t start_ofs;
|
||
int match_len;
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
ofpraw_put(OFPRAW_NXST_FLOW_MONITOR_REQUEST, OFP10_VERSION, msg);
|
||
}
|
||
|
||
start_ofs = ofpbuf_size(msg);
|
||
ofpbuf_put_zeros(msg, sizeof *nfmr);
|
||
match_len = nx_put_match(msg, &rq->match, htonll(0), htonll(0));
|
||
|
||
nfmr = ofpbuf_at_assert(msg, start_ofs, sizeof *nfmr);
|
||
nfmr->id = htonl(rq->id);
|
||
nfmr->flags = htons(rq->flags);
|
||
nfmr->out_port = htons(ofp_to_u16(rq->out_port));
|
||
nfmr->match_len = htons(match_len);
|
||
nfmr->table_id = rq->table_id;
|
||
}
|
||
|
||
/* Converts an NXST_FLOW_MONITOR reply (also known as a flow update) in 'msg'
|
||
* into an abstract ofputil_flow_update in 'update'. The caller must have
|
||
* initialized update->match to point to space allocated for a match.
|
||
*
|
||
* Uses 'ofpacts' to store the abstract OFPACT_* version of the update's
|
||
* actions (except for NXFME_ABBREV, which never includes actions). The caller
|
||
* must initialize 'ofpacts' and retains ownership of it. 'update->ofpacts'
|
||
* will point into the 'ofpacts' buffer.
|
||
*
|
||
* Multiple flow updates can be packed into a single OpenFlow message. Calling
|
||
* this function multiple times for a single 'msg' iterates through the
|
||
* updates. The caller must initially leave 'msg''s layer pointers null and
|
||
* not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no updates were left in this 'msg',
|
||
* otherwise an OFPERR_* value. */
|
||
int
|
||
ofputil_decode_flow_update(struct ofputil_flow_update *update,
|
||
struct ofpbuf *msg, struct ofpbuf *ofpacts)
|
||
{
|
||
struct nx_flow_update_header *nfuh;
|
||
unsigned int length;
|
||
struct ofp_header *oh;
|
||
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
if (ofpbuf_size(msg) < sizeof(struct nx_flow_update_header)) {
|
||
goto bad_len;
|
||
}
|
||
|
||
oh = msg->frame;
|
||
|
||
nfuh = ofpbuf_data(msg);
|
||
update->event = ntohs(nfuh->event);
|
||
length = ntohs(nfuh->length);
|
||
if (length > ofpbuf_size(msg) || length % 8) {
|
||
goto bad_len;
|
||
}
|
||
|
||
if (update->event == NXFME_ABBREV) {
|
||
struct nx_flow_update_abbrev *nfua;
|
||
|
||
if (length != sizeof *nfua) {
|
||
goto bad_len;
|
||
}
|
||
|
||
nfua = ofpbuf_pull(msg, sizeof *nfua);
|
||
update->xid = nfua->xid;
|
||
return 0;
|
||
} else if (update->event == NXFME_ADDED
|
||
|| update->event == NXFME_DELETED
|
||
|| update->event == NXFME_MODIFIED) {
|
||
struct nx_flow_update_full *nfuf;
|
||
unsigned int actions_len;
|
||
unsigned int match_len;
|
||
enum ofperr error;
|
||
|
||
if (length < sizeof *nfuf) {
|
||
goto bad_len;
|
||
}
|
||
|
||
nfuf = ofpbuf_pull(msg, sizeof *nfuf);
|
||
match_len = ntohs(nfuf->match_len);
|
||
if (sizeof *nfuf + match_len > length) {
|
||
goto bad_len;
|
||
}
|
||
|
||
update->reason = ntohs(nfuf->reason);
|
||
update->idle_timeout = ntohs(nfuf->idle_timeout);
|
||
update->hard_timeout = ntohs(nfuf->hard_timeout);
|
||
update->table_id = nfuf->table_id;
|
||
update->cookie = nfuf->cookie;
|
||
update->priority = ntohs(nfuf->priority);
|
||
|
||
error = nx_pull_match(msg, match_len, update->match, NULL, NULL);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
actions_len = length - sizeof *nfuf - ROUND_UP(match_len, 8);
|
||
error = ofpacts_pull_openflow_actions(msg, actions_len, oh->version,
|
||
ofpacts);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
update->ofpacts = ofpbuf_data(ofpacts);
|
||
update->ofpacts_len = ofpbuf_size(ofpacts);
|
||
return 0;
|
||
} else {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl,
|
||
"NXST_FLOW_MONITOR reply has bad event %"PRIu16,
|
||
ntohs(nfuh->event));
|
||
return OFPERR_NXBRC_FM_BAD_EVENT;
|
||
}
|
||
|
||
bad_len:
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "NXST_FLOW_MONITOR reply has %"PRIu32" "
|
||
"leftover bytes at end", ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
uint32_t
|
||
ofputil_decode_flow_monitor_cancel(const struct ofp_header *oh)
|
||
{
|
||
const struct nx_flow_monitor_cancel *cancel = ofpmsg_body(oh);
|
||
|
||
return ntohl(cancel->id);
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_flow_monitor_cancel(uint32_t id)
|
||
{
|
||
struct nx_flow_monitor_cancel *nfmc;
|
||
struct ofpbuf *msg;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_NXT_FLOW_MONITOR_CANCEL, OFP10_VERSION, 0);
|
||
nfmc = ofpbuf_put_uninit(msg, sizeof *nfmc);
|
||
nfmc->id = htonl(id);
|
||
return msg;
|
||
}
|
||
|
||
void
|
||
ofputil_start_flow_update(struct list *replies)
|
||
{
|
||
struct ofpbuf *msg;
|
||
|
||
msg = ofpraw_alloc_xid(OFPRAW_NXST_FLOW_MONITOR_REPLY, OFP10_VERSION,
|
||
htonl(0), 1024);
|
||
|
||
list_init(replies);
|
||
list_push_back(replies, &msg->list_node);
|
||
}
|
||
|
||
void
|
||
ofputil_append_flow_update(const struct ofputil_flow_update *update,
|
||
struct list *replies)
|
||
{
|
||
enum ofp_version version = ofpmp_version(replies);
|
||
struct nx_flow_update_header *nfuh;
|
||
struct ofpbuf *msg;
|
||
size_t start_ofs;
|
||
|
||
msg = ofpbuf_from_list(list_back(replies));
|
||
start_ofs = ofpbuf_size(msg);
|
||
|
||
if (update->event == NXFME_ABBREV) {
|
||
struct nx_flow_update_abbrev *nfua;
|
||
|
||
nfua = ofpbuf_put_zeros(msg, sizeof *nfua);
|
||
nfua->xid = update->xid;
|
||
} else {
|
||
struct nx_flow_update_full *nfuf;
|
||
int match_len;
|
||
|
||
ofpbuf_put_zeros(msg, sizeof *nfuf);
|
||
match_len = nx_put_match(msg, update->match, htonll(0), htonll(0));
|
||
ofpacts_put_openflow_actions(update->ofpacts, update->ofpacts_len, msg,
|
||
version);
|
||
nfuf = ofpbuf_at_assert(msg, start_ofs, sizeof *nfuf);
|
||
nfuf->reason = htons(update->reason);
|
||
nfuf->priority = htons(update->priority);
|
||
nfuf->idle_timeout = htons(update->idle_timeout);
|
||
nfuf->hard_timeout = htons(update->hard_timeout);
|
||
nfuf->match_len = htons(match_len);
|
||
nfuf->table_id = update->table_id;
|
||
nfuf->cookie = update->cookie;
|
||
}
|
||
|
||
nfuh = ofpbuf_at_assert(msg, start_ofs, sizeof *nfuh);
|
||
nfuh->length = htons(ofpbuf_size(msg) - start_ofs);
|
||
nfuh->event = htons(update->event);
|
||
|
||
ofpmp_postappend(replies, start_ofs);
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_packet_out(const struct ofputil_packet_out *po,
|
||
enum ofputil_protocol protocol)
|
||
{
|
||
enum ofp_version ofp_version = ofputil_protocol_to_ofp_version(protocol);
|
||
struct ofpbuf *msg;
|
||
size_t size;
|
||
|
||
size = po->ofpacts_len;
|
||
if (po->buffer_id == UINT32_MAX) {
|
||
size += po->packet_len;
|
||
}
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
struct ofp10_packet_out *opo;
|
||
size_t actions_ofs;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_OFPT10_PACKET_OUT, OFP10_VERSION, size);
|
||
ofpbuf_put_zeros(msg, sizeof *opo);
|
||
actions_ofs = ofpbuf_size(msg);
|
||
ofpacts_put_openflow_actions(po->ofpacts, po->ofpacts_len, msg,
|
||
ofp_version);
|
||
|
||
opo = ofpbuf_l3(msg);
|
||
opo->buffer_id = htonl(po->buffer_id);
|
||
opo->in_port = htons(ofp_to_u16(po->in_port));
|
||
opo->actions_len = htons(ofpbuf_size(msg) - actions_ofs);
|
||
break;
|
||
}
|
||
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp11_packet_out *opo;
|
||
size_t len;
|
||
|
||
msg = ofpraw_alloc(OFPRAW_OFPT11_PACKET_OUT, ofp_version, size);
|
||
ofpbuf_put_zeros(msg, sizeof *opo);
|
||
len = ofpacts_put_openflow_actions(po->ofpacts, po->ofpacts_len, msg,
|
||
ofp_version);
|
||
opo = ofpbuf_l3(msg);
|
||
opo->buffer_id = htonl(po->buffer_id);
|
||
opo->in_port = ofputil_port_to_ofp11(po->in_port);
|
||
opo->actions_len = htons(len);
|
||
break;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
if (po->buffer_id == UINT32_MAX) {
|
||
ofpbuf_put(msg, po->packet, po->packet_len);
|
||
}
|
||
|
||
ofpmsg_update_length(msg);
|
||
|
||
return msg;
|
||
}
|
||
|
||
/* Creates and returns an OFPT_ECHO_REQUEST message with an empty payload. */
|
||
struct ofpbuf *
|
||
make_echo_request(enum ofp_version ofp_version)
|
||
{
|
||
return ofpraw_alloc_xid(OFPRAW_OFPT_ECHO_REQUEST, ofp_version,
|
||
htonl(0), 0);
|
||
}
|
||
|
||
/* Creates and returns an OFPT_ECHO_REPLY message matching the
|
||
* OFPT_ECHO_REQUEST message in 'rq'. */
|
||
struct ofpbuf *
|
||
make_echo_reply(const struct ofp_header *rq)
|
||
{
|
||
struct ofpbuf rq_buf;
|
||
struct ofpbuf *reply;
|
||
|
||
ofpbuf_use_const(&rq_buf, rq, ntohs(rq->length));
|
||
ofpraw_pull_assert(&rq_buf);
|
||
|
||
reply = ofpraw_alloc_reply(OFPRAW_OFPT_ECHO_REPLY, rq, ofpbuf_size(&rq_buf));
|
||
ofpbuf_put(reply, ofpbuf_data(&rq_buf), ofpbuf_size(&rq_buf));
|
||
return reply;
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_barrier_request(enum ofp_version ofp_version)
|
||
{
|
||
enum ofpraw type;
|
||
|
||
switch (ofp_version) {
|
||
case OFP15_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP11_VERSION:
|
||
type = OFPRAW_OFPT11_BARRIER_REQUEST;
|
||
break;
|
||
|
||
case OFP10_VERSION:
|
||
type = OFPRAW_OFPT10_BARRIER_REQUEST;
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return ofpraw_alloc(type, ofp_version, 0);
|
||
}
|
||
|
||
const char *
|
||
ofputil_frag_handling_to_string(enum ofp_config_flags flags)
|
||
{
|
||
switch (flags & OFPC_FRAG_MASK) {
|
||
case OFPC_FRAG_NORMAL: return "normal";
|
||
case OFPC_FRAG_DROP: return "drop";
|
||
case OFPC_FRAG_REASM: return "reassemble";
|
||
case OFPC_FRAG_NX_MATCH: return "nx-match";
|
||
}
|
||
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
bool
|
||
ofputil_frag_handling_from_string(const char *s, enum ofp_config_flags *flags)
|
||
{
|
||
if (!strcasecmp(s, "normal")) {
|
||
*flags = OFPC_FRAG_NORMAL;
|
||
} else if (!strcasecmp(s, "drop")) {
|
||
*flags = OFPC_FRAG_DROP;
|
||
} else if (!strcasecmp(s, "reassemble")) {
|
||
*flags = OFPC_FRAG_REASM;
|
||
} else if (!strcasecmp(s, "nx-match")) {
|
||
*flags = OFPC_FRAG_NX_MATCH;
|
||
} else {
|
||
return false;
|
||
}
|
||
return true;
|
||
}
|
||
|
||
/* Converts the OpenFlow 1.1+ port number 'ofp11_port' into an OpenFlow 1.0
|
||
* port number and stores the latter in '*ofp10_port', for the purpose of
|
||
* decoding OpenFlow 1.1+ protocol messages. Returns 0 if successful,
|
||
* otherwise an OFPERR_* number. On error, stores OFPP_NONE in '*ofp10_port'.
|
||
*
|
||
* See the definition of OFP11_MAX for an explanation of the mapping. */
|
||
enum ofperr
|
||
ofputil_port_from_ofp11(ovs_be32 ofp11_port, ofp_port_t *ofp10_port)
|
||
{
|
||
uint32_t ofp11_port_h = ntohl(ofp11_port);
|
||
|
||
if (ofp11_port_h < ofp_to_u16(OFPP_MAX)) {
|
||
*ofp10_port = u16_to_ofp(ofp11_port_h);
|
||
return 0;
|
||
} else if (ofp11_port_h >= ofp11_to_u32(OFPP11_MAX)) {
|
||
*ofp10_port = u16_to_ofp(ofp11_port_h - OFPP11_OFFSET);
|
||
return 0;
|
||
} else {
|
||
*ofp10_port = OFPP_NONE;
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "port %"PRIu32" is outside the supported "
|
||
"range 0 through %d or 0x%"PRIx32" through 0x%"PRIx32,
|
||
ofp11_port_h, ofp_to_u16(OFPP_MAX) - 1,
|
||
ofp11_to_u32(OFPP11_MAX), UINT32_MAX);
|
||
return OFPERR_OFPBAC_BAD_OUT_PORT;
|
||
}
|
||
}
|
||
|
||
/* Returns the OpenFlow 1.1+ port number equivalent to the OpenFlow 1.0 port
|
||
* number 'ofp10_port', for encoding OpenFlow 1.1+ protocol messages.
|
||
*
|
||
* See the definition of OFP11_MAX for an explanation of the mapping. */
|
||
ovs_be32
|
||
ofputil_port_to_ofp11(ofp_port_t ofp10_port)
|
||
{
|
||
return htonl(ofp_to_u16(ofp10_port) < ofp_to_u16(OFPP_MAX)
|
||
? ofp_to_u16(ofp10_port)
|
||
: ofp_to_u16(ofp10_port) + OFPP11_OFFSET);
|
||
}
|
||
|
||
#define OFPUTIL_NAMED_PORTS \
|
||
OFPUTIL_NAMED_PORT(IN_PORT) \
|
||
OFPUTIL_NAMED_PORT(TABLE) \
|
||
OFPUTIL_NAMED_PORT(NORMAL) \
|
||
OFPUTIL_NAMED_PORT(FLOOD) \
|
||
OFPUTIL_NAMED_PORT(ALL) \
|
||
OFPUTIL_NAMED_PORT(CONTROLLER) \
|
||
OFPUTIL_NAMED_PORT(LOCAL) \
|
||
OFPUTIL_NAMED_PORT(ANY)
|
||
|
||
/* For backwards compatibility, so that "none" is recognized as OFPP_ANY */
|
||
#define OFPUTIL_NAMED_PORTS_WITH_NONE \
|
||
OFPUTIL_NAMED_PORTS \
|
||
OFPUTIL_NAMED_PORT(NONE)
|
||
|
||
/* Stores the port number represented by 's' into '*portp'. 's' may be an
|
||
* integer or, for reserved ports, the standard OpenFlow name for the port
|
||
* (e.g. "LOCAL").
|
||
*
|
||
* Returns true if successful, false if 's' is not a valid OpenFlow port number
|
||
* or name. The caller should issue an error message in this case, because
|
||
* this function usually does not. (This gives the caller an opportunity to
|
||
* look up the port name another way, e.g. by contacting the switch and listing
|
||
* the names of all its ports).
|
||
*
|
||
* This function accepts OpenFlow 1.0 port numbers. It also accepts a subset
|
||
* of OpenFlow 1.1+ port numbers, mapping those port numbers into the 16-bit
|
||
* range as described in include/openflow/openflow-1.1.h. */
|
||
bool
|
||
ofputil_port_from_string(const char *s, ofp_port_t *portp)
|
||
{
|
||
unsigned int port32; /* int is at least 32 bits wide. */
|
||
|
||
if (*s == '-') {
|
||
VLOG_WARN("Negative value %s is not a valid port number.", s);
|
||
return false;
|
||
}
|
||
*portp = 0;
|
||
if (str_to_uint(s, 10, &port32)) {
|
||
if (port32 < ofp_to_u16(OFPP_MAX)) {
|
||
/* Pass. */
|
||
} else if (port32 < ofp_to_u16(OFPP_FIRST_RESV)) {
|
||
VLOG_WARN("port %u is a reserved OF1.0 port number that will "
|
||
"be translated to %u when talking to an OF1.1 or "
|
||
"later controller", port32, port32 + OFPP11_OFFSET);
|
||
} else if (port32 <= ofp_to_u16(OFPP_LAST_RESV)) {
|
||
char name[OFP_MAX_PORT_NAME_LEN];
|
||
|
||
ofputil_port_to_string(u16_to_ofp(port32), name, sizeof name);
|
||
VLOG_WARN_ONCE("referring to port %s as %"PRIu32" is deprecated "
|
||
"for compatibility with OpenFlow 1.1 and later",
|
||
name, port32);
|
||
} else if (port32 < ofp11_to_u32(OFPP11_MAX)) {
|
||
VLOG_WARN("port %u is outside the supported range 0 through "
|
||
"%"PRIx16" or 0x%x through 0x%"PRIx32, port32,
|
||
UINT16_MAX, ofp11_to_u32(OFPP11_MAX), UINT32_MAX);
|
||
return false;
|
||
} else {
|
||
port32 -= OFPP11_OFFSET;
|
||
}
|
||
|
||
*portp = u16_to_ofp(port32);
|
||
return true;
|
||
} else {
|
||
struct pair {
|
||
const char *name;
|
||
ofp_port_t value;
|
||
};
|
||
static const struct pair pairs[] = {
|
||
#define OFPUTIL_NAMED_PORT(NAME) {#NAME, OFPP_##NAME},
|
||
OFPUTIL_NAMED_PORTS_WITH_NONE
|
||
#undef OFPUTIL_NAMED_PORT
|
||
};
|
||
const struct pair *p;
|
||
|
||
for (p = pairs; p < &pairs[ARRAY_SIZE(pairs)]; p++) {
|
||
if (!strcasecmp(s, p->name)) {
|
||
*portp = p->value;
|
||
return true;
|
||
}
|
||
}
|
||
return false;
|
||
}
|
||
}
|
||
|
||
/* Appends to 's' a string representation of the OpenFlow port number 'port'.
|
||
* Most ports' string representation is just the port number, but for special
|
||
* ports, e.g. OFPP_LOCAL, it is the name, e.g. "LOCAL". */
|
||
void
|
||
ofputil_format_port(ofp_port_t port, struct ds *s)
|
||
{
|
||
char name[OFP_MAX_PORT_NAME_LEN];
|
||
|
||
ofputil_port_to_string(port, name, sizeof name);
|
||
ds_put_cstr(s, name);
|
||
}
|
||
|
||
/* Puts in the 'bufsize' byte in 'namebuf' a null-terminated string
|
||
* representation of OpenFlow port number 'port'. Most ports are represented
|
||
* as just the port number, but special ports, e.g. OFPP_LOCAL, are represented
|
||
* by name, e.g. "LOCAL". */
|
||
void
|
||
ofputil_port_to_string(ofp_port_t port,
|
||
char namebuf[OFP_MAX_PORT_NAME_LEN], size_t bufsize)
|
||
{
|
||
switch (port) {
|
||
#define OFPUTIL_NAMED_PORT(NAME) \
|
||
case OFPP_##NAME: \
|
||
ovs_strlcpy(namebuf, #NAME, bufsize); \
|
||
break;
|
||
OFPUTIL_NAMED_PORTS
|
||
#undef OFPUTIL_NAMED_PORT
|
||
|
||
default:
|
||
snprintf(namebuf, bufsize, "%"PRIu16, port);
|
||
break;
|
||
}
|
||
}
|
||
|
||
/* Stores the group id represented by 's' into '*group_idp'. 's' may be an
|
||
* integer or, for reserved group IDs, the standard OpenFlow name for the group
|
||
* (either "ANY" or "ALL").
|
||
*
|
||
* Returns true if successful, false if 's' is not a valid OpenFlow group ID or
|
||
* name. */
|
||
bool
|
||
ofputil_group_from_string(const char *s, uint32_t *group_idp)
|
||
{
|
||
if (!strcasecmp(s, "any")) {
|
||
*group_idp = OFPG11_ANY;
|
||
} else if (!strcasecmp(s, "all")) {
|
||
*group_idp = OFPG11_ALL;
|
||
} else if (!str_to_uint(s, 10, group_idp)) {
|
||
VLOG_WARN("%s is not a valid group ID. (Valid group IDs are "
|
||
"32-bit nonnegative integers or the keywords ANY or "
|
||
"ALL.)", s);
|
||
return false;
|
||
}
|
||
|
||
return true;
|
||
}
|
||
|
||
/* Appends to 's' a string representation of the OpenFlow group ID 'group_id'.
|
||
* Most groups' string representation is just the number, but for special
|
||
* groups, e.g. OFPG11_ALL, it is the name, e.g. "ALL". */
|
||
void
|
||
ofputil_format_group(uint32_t group_id, struct ds *s)
|
||
{
|
||
char name[MAX_GROUP_NAME_LEN];
|
||
|
||
ofputil_group_to_string(group_id, name, sizeof name);
|
||
ds_put_cstr(s, name);
|
||
}
|
||
|
||
|
||
/* Puts in the 'bufsize' byte in 'namebuf' a null-terminated string
|
||
* representation of OpenFlow group ID 'group_id'. Most group are represented
|
||
* as just their number, but special groups, e.g. OFPG11_ALL, are represented
|
||
* by name, e.g. "ALL". */
|
||
void
|
||
ofputil_group_to_string(uint32_t group_id,
|
||
char namebuf[MAX_GROUP_NAME_LEN + 1], size_t bufsize)
|
||
{
|
||
switch (group_id) {
|
||
case OFPG11_ALL:
|
||
ovs_strlcpy(namebuf, "ALL", bufsize);
|
||
break;
|
||
|
||
case OFPG11_ANY:
|
||
ovs_strlcpy(namebuf, "ANY", bufsize);
|
||
break;
|
||
|
||
default:
|
||
snprintf(namebuf, bufsize, "%"PRIu32, group_id);
|
||
break;
|
||
}
|
||
}
|
||
|
||
/* Given a buffer 'b' that contains an array of OpenFlow ports of type
|
||
* 'ofp_version', tries to pull the first element from the array. If
|
||
* successful, initializes '*pp' with an abstract representation of the
|
||
* port and returns 0. If no ports remain to be decoded, returns EOF.
|
||
* On an error, returns a positive OFPERR_* value. */
|
||
int
|
||
ofputil_pull_phy_port(enum ofp_version ofp_version, struct ofpbuf *b,
|
||
struct ofputil_phy_port *pp)
|
||
{
|
||
memset(pp, 0, sizeof *pp);
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
const struct ofp10_phy_port *opp = ofpbuf_try_pull(b, sizeof *opp);
|
||
return opp ? ofputil_decode_ofp10_phy_port(pp, opp) : EOF;
|
||
}
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION: {
|
||
const struct ofp11_port *op = ofpbuf_try_pull(b, sizeof *op);
|
||
return op ? ofputil_decode_ofp11_port(pp, op) : EOF;
|
||
}
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
return ofpbuf_size(b) ? ofputil_pull_ofp14_port(pp, b) : EOF;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
static void
|
||
ofputil_normalize_match__(struct match *match, bool may_log)
|
||
{
|
||
enum {
|
||
MAY_NW_ADDR = 1 << 0, /* nw_src, nw_dst */
|
||
MAY_TP_ADDR = 1 << 1, /* tp_src, tp_dst */
|
||
MAY_NW_PROTO = 1 << 2, /* nw_proto */
|
||
MAY_IPVx = 1 << 3, /* tos, frag, ttl */
|
||
MAY_ARP_SHA = 1 << 4, /* arp_sha */
|
||
MAY_ARP_THA = 1 << 5, /* arp_tha */
|
||
MAY_IPV6 = 1 << 6, /* ipv6_src, ipv6_dst, ipv6_label */
|
||
MAY_ND_TARGET = 1 << 7, /* nd_target */
|
||
MAY_MPLS = 1 << 8, /* mpls label and tc */
|
||
} may_match;
|
||
|
||
struct flow_wildcards wc;
|
||
|
||
/* Figure out what fields may be matched. */
|
||
if (match->flow.dl_type == htons(ETH_TYPE_IP)) {
|
||
may_match = MAY_NW_PROTO | MAY_IPVx | MAY_NW_ADDR;
|
||
if (match->flow.nw_proto == IPPROTO_TCP ||
|
||
match->flow.nw_proto == IPPROTO_UDP ||
|
||
match->flow.nw_proto == IPPROTO_SCTP ||
|
||
match->flow.nw_proto == IPPROTO_ICMP) {
|
||
may_match |= MAY_TP_ADDR;
|
||
}
|
||
} else if (match->flow.dl_type == htons(ETH_TYPE_IPV6)) {
|
||
may_match = MAY_NW_PROTO | MAY_IPVx | MAY_IPV6;
|
||
if (match->flow.nw_proto == IPPROTO_TCP ||
|
||
match->flow.nw_proto == IPPROTO_UDP ||
|
||
match->flow.nw_proto == IPPROTO_SCTP) {
|
||
may_match |= MAY_TP_ADDR;
|
||
} else if (match->flow.nw_proto == IPPROTO_ICMPV6) {
|
||
may_match |= MAY_TP_ADDR;
|
||
if (match->flow.tp_src == htons(ND_NEIGHBOR_SOLICIT)) {
|
||
may_match |= MAY_ND_TARGET | MAY_ARP_SHA;
|
||
} else if (match->flow.tp_src == htons(ND_NEIGHBOR_ADVERT)) {
|
||
may_match |= MAY_ND_TARGET | MAY_ARP_THA;
|
||
}
|
||
}
|
||
} else if (match->flow.dl_type == htons(ETH_TYPE_ARP) ||
|
||
match->flow.dl_type == htons(ETH_TYPE_RARP)) {
|
||
may_match = MAY_NW_PROTO | MAY_NW_ADDR | MAY_ARP_SHA | MAY_ARP_THA;
|
||
} else if (eth_type_mpls(match->flow.dl_type)) {
|
||
may_match = MAY_MPLS;
|
||
} else {
|
||
may_match = 0;
|
||
}
|
||
|
||
/* Clear the fields that may not be matched. */
|
||
wc = match->wc;
|
||
if (!(may_match & MAY_NW_ADDR)) {
|
||
wc.masks.nw_src = wc.masks.nw_dst = htonl(0);
|
||
}
|
||
if (!(may_match & MAY_TP_ADDR)) {
|
||
wc.masks.tp_src = wc.masks.tp_dst = htons(0);
|
||
}
|
||
if (!(may_match & MAY_NW_PROTO)) {
|
||
wc.masks.nw_proto = 0;
|
||
}
|
||
if (!(may_match & MAY_IPVx)) {
|
||
wc.masks.nw_tos = 0;
|
||
wc.masks.nw_ttl = 0;
|
||
}
|
||
if (!(may_match & MAY_ARP_SHA)) {
|
||
memset(wc.masks.arp_sha, 0, ETH_ADDR_LEN);
|
||
}
|
||
if (!(may_match & MAY_ARP_THA)) {
|
||
memset(wc.masks.arp_tha, 0, ETH_ADDR_LEN);
|
||
}
|
||
if (!(may_match & MAY_IPV6)) {
|
||
wc.masks.ipv6_src = wc.masks.ipv6_dst = in6addr_any;
|
||
wc.masks.ipv6_label = htonl(0);
|
||
}
|
||
if (!(may_match & MAY_ND_TARGET)) {
|
||
wc.masks.nd_target = in6addr_any;
|
||
}
|
||
if (!(may_match & MAY_MPLS)) {
|
||
memset(wc.masks.mpls_lse, 0, sizeof wc.masks.mpls_lse);
|
||
}
|
||
|
||
/* Log any changes. */
|
||
if (!flow_wildcards_equal(&wc, &match->wc)) {
|
||
bool log = may_log && !VLOG_DROP_INFO(&bad_ofmsg_rl);
|
||
char *pre = log ? match_to_string(match, OFP_DEFAULT_PRIORITY) : NULL;
|
||
|
||
match->wc = wc;
|
||
match_zero_wildcarded_fields(match);
|
||
|
||
if (log) {
|
||
char *post = match_to_string(match, OFP_DEFAULT_PRIORITY);
|
||
VLOG_INFO("normalization changed ofp_match, details:");
|
||
VLOG_INFO(" pre: %s", pre);
|
||
VLOG_INFO("post: %s", post);
|
||
free(pre);
|
||
free(post);
|
||
}
|
||
}
|
||
}
|
||
|
||
/* "Normalizes" the wildcards in 'match'. That means:
|
||
*
|
||
* 1. If the type of level N is known, then only the valid fields for that
|
||
* level may be specified. For example, ARP does not have a TOS field,
|
||
* so nw_tos must be wildcarded if 'match' specifies an ARP flow.
|
||
* Similarly, IPv4 does not have any IPv6 addresses, so ipv6_src and
|
||
* ipv6_dst (and other fields) must be wildcarded if 'match' specifies an
|
||
* IPv4 flow.
|
||
*
|
||
* 2. If the type of level N is not known (or not understood by Open
|
||
* vSwitch), then no fields at all for that level may be specified. For
|
||
* example, Open vSwitch does not understand SCTP, an L4 protocol, so the
|
||
* L4 fields tp_src and tp_dst must be wildcarded if 'match' specifies an
|
||
* SCTP flow.
|
||
*
|
||
* If this function changes 'match', it logs a rate-limited informational
|
||
* message. */
|
||
void
|
||
ofputil_normalize_match(struct match *match)
|
||
{
|
||
ofputil_normalize_match__(match, true);
|
||
}
|
||
|
||
/* Same as ofputil_normalize_match() without the logging. Thus, this function
|
||
* is suitable for a program's internal use, whereas ofputil_normalize_match()
|
||
* sense for use on flows received from elsewhere (so that a bug in the program
|
||
* that sent them can be reported and corrected). */
|
||
void
|
||
ofputil_normalize_match_quiet(struct match *match)
|
||
{
|
||
ofputil_normalize_match__(match, false);
|
||
}
|
||
|
||
/* Parses a key or a key-value pair from '*stringp'.
|
||
*
|
||
* On success: Stores the key into '*keyp'. Stores the value, if present, into
|
||
* '*valuep', otherwise an empty string. Advances '*stringp' past the end of
|
||
* the key-value pair, preparing it for another call. '*keyp' and '*valuep'
|
||
* are substrings of '*stringp' created by replacing some of its bytes by null
|
||
* terminators. Returns true.
|
||
*
|
||
* If '*stringp' is just white space or commas, sets '*keyp' and '*valuep' to
|
||
* NULL and returns false. */
|
||
bool
|
||
ofputil_parse_key_value(char **stringp, char **keyp, char **valuep)
|
||
{
|
||
char *pos, *key, *value;
|
||
size_t key_len;
|
||
|
||
pos = *stringp;
|
||
pos += strspn(pos, ", \t\r\n");
|
||
if (*pos == '\0') {
|
||
*keyp = *valuep = NULL;
|
||
return false;
|
||
}
|
||
|
||
key = pos;
|
||
key_len = strcspn(pos, ":=(, \t\r\n");
|
||
if (key[key_len] == ':' || key[key_len] == '=') {
|
||
/* The value can be separated by a colon. */
|
||
size_t value_len;
|
||
|
||
value = key + key_len + 1;
|
||
value_len = strcspn(value, ", \t\r\n");
|
||
pos = value + value_len + (value[value_len] != '\0');
|
||
value[value_len] = '\0';
|
||
} else if (key[key_len] == '(') {
|
||
/* The value can be surrounded by balanced parentheses. The outermost
|
||
* set of parentheses is removed. */
|
||
int level = 1;
|
||
size_t value_len;
|
||
|
||
value = key + key_len + 1;
|
||
for (value_len = 0; level > 0; value_len++) {
|
||
switch (value[value_len]) {
|
||
case '\0':
|
||
level = 0;
|
||
break;
|
||
|
||
case '(':
|
||
level++;
|
||
break;
|
||
|
||
case ')':
|
||
level--;
|
||
break;
|
||
}
|
||
}
|
||
value[value_len - 1] = '\0';
|
||
pos = value + value_len;
|
||
} else {
|
||
/* There might be no value at all. */
|
||
value = key + key_len; /* Will become the empty string below. */
|
||
pos = key + key_len + (key[key_len] != '\0');
|
||
}
|
||
key[key_len] = '\0';
|
||
|
||
*stringp = pos;
|
||
*keyp = key;
|
||
*valuep = value;
|
||
return true;
|
||
}
|
||
|
||
/* Encode a dump ports request for 'port', the encoded message
|
||
* will be for Open Flow version 'ofp_version'. Returns message
|
||
* as a struct ofpbuf. Returns encoded message on success, NULL on error */
|
||
struct ofpbuf *
|
||
ofputil_encode_dump_ports_request(enum ofp_version ofp_version, ofp_port_t port)
|
||
{
|
||
struct ofpbuf *request;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
struct ofp10_port_stats_request *req;
|
||
request = ofpraw_alloc(OFPRAW_OFPST10_PORT_REQUEST, ofp_version, 0);
|
||
req = ofpbuf_put_zeros(request, sizeof *req);
|
||
req->port_no = htons(ofp_to_u16(port));
|
||
break;
|
||
}
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp11_port_stats_request *req;
|
||
request = ofpraw_alloc(OFPRAW_OFPST11_PORT_REQUEST, ofp_version, 0);
|
||
req = ofpbuf_put_zeros(request, sizeof *req);
|
||
req->port_no = ofputil_port_to_ofp11(port);
|
||
break;
|
||
}
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
static void
|
||
ofputil_port_stats_to_ofp10(const struct ofputil_port_stats *ops,
|
||
struct ofp10_port_stats *ps10)
|
||
{
|
||
ps10->port_no = htons(ofp_to_u16(ops->port_no));
|
||
memset(ps10->pad, 0, sizeof ps10->pad);
|
||
put_32aligned_be64(&ps10->rx_packets, htonll(ops->stats.rx_packets));
|
||
put_32aligned_be64(&ps10->tx_packets, htonll(ops->stats.tx_packets));
|
||
put_32aligned_be64(&ps10->rx_bytes, htonll(ops->stats.rx_bytes));
|
||
put_32aligned_be64(&ps10->tx_bytes, htonll(ops->stats.tx_bytes));
|
||
put_32aligned_be64(&ps10->rx_dropped, htonll(ops->stats.rx_dropped));
|
||
put_32aligned_be64(&ps10->tx_dropped, htonll(ops->stats.tx_dropped));
|
||
put_32aligned_be64(&ps10->rx_errors, htonll(ops->stats.rx_errors));
|
||
put_32aligned_be64(&ps10->tx_errors, htonll(ops->stats.tx_errors));
|
||
put_32aligned_be64(&ps10->rx_frame_err, htonll(ops->stats.rx_frame_errors));
|
||
put_32aligned_be64(&ps10->rx_over_err, htonll(ops->stats.rx_over_errors));
|
||
put_32aligned_be64(&ps10->rx_crc_err, htonll(ops->stats.rx_crc_errors));
|
||
put_32aligned_be64(&ps10->collisions, htonll(ops->stats.collisions));
|
||
}
|
||
|
||
static void
|
||
ofputil_port_stats_to_ofp11(const struct ofputil_port_stats *ops,
|
||
struct ofp11_port_stats *ps11)
|
||
{
|
||
ps11->port_no = ofputil_port_to_ofp11(ops->port_no);
|
||
memset(ps11->pad, 0, sizeof ps11->pad);
|
||
ps11->rx_packets = htonll(ops->stats.rx_packets);
|
||
ps11->tx_packets = htonll(ops->stats.tx_packets);
|
||
ps11->rx_bytes = htonll(ops->stats.rx_bytes);
|
||
ps11->tx_bytes = htonll(ops->stats.tx_bytes);
|
||
ps11->rx_dropped = htonll(ops->stats.rx_dropped);
|
||
ps11->tx_dropped = htonll(ops->stats.tx_dropped);
|
||
ps11->rx_errors = htonll(ops->stats.rx_errors);
|
||
ps11->tx_errors = htonll(ops->stats.tx_errors);
|
||
ps11->rx_frame_err = htonll(ops->stats.rx_frame_errors);
|
||
ps11->rx_over_err = htonll(ops->stats.rx_over_errors);
|
||
ps11->rx_crc_err = htonll(ops->stats.rx_crc_errors);
|
||
ps11->collisions = htonll(ops->stats.collisions);
|
||
}
|
||
|
||
static void
|
||
ofputil_port_stats_to_ofp13(const struct ofputil_port_stats *ops,
|
||
struct ofp13_port_stats *ps13)
|
||
{
|
||
ofputil_port_stats_to_ofp11(ops, &ps13->ps);
|
||
ps13->duration_sec = htonl(ops->duration_sec);
|
||
ps13->duration_nsec = htonl(ops->duration_nsec);
|
||
}
|
||
|
||
static void
|
||
ofputil_append_ofp14_port_stats(const struct ofputil_port_stats *ops,
|
||
struct list *replies)
|
||
{
|
||
struct ofp14_port_stats_prop_ethernet *eth;
|
||
struct ofp14_port_stats *ps14;
|
||
struct ofpbuf *reply;
|
||
|
||
reply = ofpmp_reserve(replies, sizeof *ps14 + sizeof *eth);
|
||
|
||
ps14 = ofpbuf_put_uninit(reply, sizeof *ps14);
|
||
ps14->length = htons(sizeof *ps14 + sizeof *eth);
|
||
memset(ps14->pad, 0, sizeof ps14->pad);
|
||
ps14->port_no = ofputil_port_to_ofp11(ops->port_no);
|
||
ps14->duration_sec = htonl(ops->duration_sec);
|
||
ps14->duration_nsec = htonl(ops->duration_nsec);
|
||
ps14->rx_packets = htonll(ops->stats.rx_packets);
|
||
ps14->tx_packets = htonll(ops->stats.tx_packets);
|
||
ps14->rx_bytes = htonll(ops->stats.rx_bytes);
|
||
ps14->tx_bytes = htonll(ops->stats.tx_bytes);
|
||
ps14->rx_dropped = htonll(ops->stats.rx_dropped);
|
||
ps14->tx_dropped = htonll(ops->stats.tx_dropped);
|
||
ps14->rx_errors = htonll(ops->stats.rx_errors);
|
||
ps14->tx_errors = htonll(ops->stats.tx_errors);
|
||
|
||
eth = ofpbuf_put_uninit(reply, sizeof *eth);
|
||
eth->type = htons(OFPPSPT14_ETHERNET);
|
||
eth->length = htons(sizeof *eth);
|
||
memset(eth->pad, 0, sizeof eth->pad);
|
||
eth->rx_frame_err = htonll(ops->stats.rx_frame_errors);
|
||
eth->rx_over_err = htonll(ops->stats.rx_over_errors);
|
||
eth->rx_crc_err = htonll(ops->stats.rx_crc_errors);
|
||
eth->collisions = htonll(ops->stats.collisions);
|
||
}
|
||
|
||
/* Encode a ports stat for 'ops' and append it to 'replies'. */
|
||
void
|
||
ofputil_append_port_stat(struct list *replies,
|
||
const struct ofputil_port_stats *ops)
|
||
{
|
||
switch (ofpmp_version(replies)) {
|
||
case OFP13_VERSION: {
|
||
struct ofp13_port_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_port_stats_to_ofp13(ops, reply);
|
||
break;
|
||
}
|
||
case OFP12_VERSION:
|
||
case OFP11_VERSION: {
|
||
struct ofp11_port_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_port_stats_to_ofp11(ops, reply);
|
||
break;
|
||
}
|
||
|
||
case OFP10_VERSION: {
|
||
struct ofp10_port_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_port_stats_to_ofp10(ops, reply);
|
||
break;
|
||
}
|
||
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
ofputil_append_ofp14_port_stats(ops, replies);
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_port_stats_from_ofp10(struct ofputil_port_stats *ops,
|
||
const struct ofp10_port_stats *ps10)
|
||
{
|
||
memset(ops, 0, sizeof *ops);
|
||
|
||
ops->port_no = u16_to_ofp(ntohs(ps10->port_no));
|
||
ops->stats.rx_packets = ntohll(get_32aligned_be64(&ps10->rx_packets));
|
||
ops->stats.tx_packets = ntohll(get_32aligned_be64(&ps10->tx_packets));
|
||
ops->stats.rx_bytes = ntohll(get_32aligned_be64(&ps10->rx_bytes));
|
||
ops->stats.tx_bytes = ntohll(get_32aligned_be64(&ps10->tx_bytes));
|
||
ops->stats.rx_dropped = ntohll(get_32aligned_be64(&ps10->rx_dropped));
|
||
ops->stats.tx_dropped = ntohll(get_32aligned_be64(&ps10->tx_dropped));
|
||
ops->stats.rx_errors = ntohll(get_32aligned_be64(&ps10->rx_errors));
|
||
ops->stats.tx_errors = ntohll(get_32aligned_be64(&ps10->tx_errors));
|
||
ops->stats.rx_frame_errors =
|
||
ntohll(get_32aligned_be64(&ps10->rx_frame_err));
|
||
ops->stats.rx_over_errors = ntohll(get_32aligned_be64(&ps10->rx_over_err));
|
||
ops->stats.rx_crc_errors = ntohll(get_32aligned_be64(&ps10->rx_crc_err));
|
||
ops->stats.collisions = ntohll(get_32aligned_be64(&ps10->collisions));
|
||
ops->duration_sec = ops->duration_nsec = UINT32_MAX;
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_port_stats_from_ofp11(struct ofputil_port_stats *ops,
|
||
const struct ofp11_port_stats *ps11)
|
||
{
|
||
enum ofperr error;
|
||
|
||
memset(ops, 0, sizeof *ops);
|
||
error = ofputil_port_from_ofp11(ps11->port_no, &ops->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
ops->stats.rx_packets = ntohll(ps11->rx_packets);
|
||
ops->stats.tx_packets = ntohll(ps11->tx_packets);
|
||
ops->stats.rx_bytes = ntohll(ps11->rx_bytes);
|
||
ops->stats.tx_bytes = ntohll(ps11->tx_bytes);
|
||
ops->stats.rx_dropped = ntohll(ps11->rx_dropped);
|
||
ops->stats.tx_dropped = ntohll(ps11->tx_dropped);
|
||
ops->stats.rx_errors = ntohll(ps11->rx_errors);
|
||
ops->stats.tx_errors = ntohll(ps11->tx_errors);
|
||
ops->stats.rx_frame_errors = ntohll(ps11->rx_frame_err);
|
||
ops->stats.rx_over_errors = ntohll(ps11->rx_over_err);
|
||
ops->stats.rx_crc_errors = ntohll(ps11->rx_crc_err);
|
||
ops->stats.collisions = ntohll(ps11->collisions);
|
||
ops->duration_sec = ops->duration_nsec = UINT32_MAX;
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_port_stats_from_ofp13(struct ofputil_port_stats *ops,
|
||
const struct ofp13_port_stats *ps13)
|
||
{
|
||
enum ofperr error = ofputil_port_stats_from_ofp11(ops, &ps13->ps);
|
||
if (!error) {
|
||
ops->duration_sec = ntohl(ps13->duration_sec);
|
||
ops->duration_nsec = ntohl(ps13->duration_nsec);
|
||
}
|
||
return error;
|
||
}
|
||
|
||
static enum ofperr
|
||
parse_ofp14_port_stats_ethernet_property(const struct ofpbuf *payload,
|
||
struct ofputil_port_stats *ops)
|
||
{
|
||
const struct ofp14_port_stats_prop_ethernet *eth = ofpbuf_data(payload);
|
||
|
||
if (ofpbuf_size(payload) != sizeof *eth) {
|
||
return OFPERR_OFPBPC_BAD_LEN;
|
||
}
|
||
|
||
ops->stats.rx_frame_errors = ntohll(eth->rx_frame_err);
|
||
ops->stats.rx_over_errors = ntohll(eth->rx_over_err);
|
||
ops->stats.rx_crc_errors = ntohll(eth->rx_crc_err);
|
||
ops->stats.collisions = ntohll(eth->collisions);
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_ofp14_port_stats(struct ofputil_port_stats *ops,
|
||
struct ofpbuf *msg)
|
||
{
|
||
const struct ofp14_port_stats *ps14;
|
||
struct ofpbuf properties;
|
||
enum ofperr error;
|
||
size_t len;
|
||
|
||
ps14 = ofpbuf_try_pull(msg, sizeof *ps14);
|
||
if (!ps14) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
len = ntohs(ps14->length);
|
||
if (len < sizeof *ps14 || len - sizeof *ps14 > ofpbuf_size(msg)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
len -= sizeof *ps14;
|
||
ofpbuf_use_const(&properties, ofpbuf_pull(msg, len), len);
|
||
|
||
error = ofputil_port_from_ofp11(ps14->port_no, &ops->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
ops->duration_sec = ntohl(ps14->duration_sec);
|
||
ops->duration_nsec = ntohl(ps14->duration_nsec);
|
||
ops->stats.rx_packets = ntohll(ps14->rx_packets);
|
||
ops->stats.tx_packets = ntohll(ps14->tx_packets);
|
||
ops->stats.rx_bytes = ntohll(ps14->rx_bytes);
|
||
ops->stats.tx_bytes = ntohll(ps14->tx_bytes);
|
||
ops->stats.rx_dropped = ntohll(ps14->rx_dropped);
|
||
ops->stats.tx_dropped = ntohll(ps14->tx_dropped);
|
||
ops->stats.rx_errors = ntohll(ps14->rx_errors);
|
||
ops->stats.tx_errors = ntohll(ps14->tx_errors);
|
||
ops->stats.rx_frame_errors = UINT64_MAX;
|
||
ops->stats.rx_over_errors = UINT64_MAX;
|
||
ops->stats.rx_crc_errors = UINT64_MAX;
|
||
ops->stats.collisions = UINT64_MAX;
|
||
|
||
while (ofpbuf_size(&properties) > 0) {
|
||
struct ofpbuf payload;
|
||
enum ofperr error;
|
||
uint16_t type;
|
||
|
||
error = ofputil_pull_property(&properties, &payload, &type);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
switch (type) {
|
||
case OFPPSPT14_ETHERNET:
|
||
error = parse_ofp14_port_stats_ethernet_property(&payload, ops);
|
||
break;
|
||
|
||
default:
|
||
log_property(true, "unknown port stats property %"PRIu16, type);
|
||
error = 0;
|
||
break;
|
||
}
|
||
|
||
if (error) {
|
||
return error;
|
||
}
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Returns the number of port stats elements in OFPTYPE_PORT_STATS_REPLY
|
||
* message 'oh'. */
|
||
size_t
|
||
ofputil_count_port_stats(const struct ofp_header *oh)
|
||
{
|
||
struct ofputil_port_stats ps;
|
||
struct ofpbuf b;
|
||
size_t n = 0;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
ofpraw_pull_assert(&b);
|
||
while (!ofputil_decode_port_stats(&ps, &b)) {
|
||
n++;
|
||
}
|
||
return n;
|
||
}
|
||
|
||
/* Converts an OFPST_PORT_STATS reply in 'msg' into an abstract
|
||
* ofputil_port_stats in 'ps'.
|
||
*
|
||
* Multiple OFPST_PORT_STATS replies can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the replies. The caller must initially leave 'msg''s layer pointers
|
||
* null and not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_port_stats(struct ofputil_port_stats *ps, struct ofpbuf *msg)
|
||
{
|
||
enum ofperr error;
|
||
enum ofpraw raw;
|
||
|
||
error = (msg->frame
|
||
? ofpraw_decode(&raw, msg->frame)
|
||
: ofpraw_pull(&raw, msg));
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
} else if (raw == OFPRAW_OFPST14_PORT_REPLY) {
|
||
return ofputil_pull_ofp14_port_stats(ps, msg);
|
||
} else if (raw == OFPRAW_OFPST13_PORT_REPLY) {
|
||
const struct ofp13_port_stats *ps13;
|
||
|
||
ps13 = ofpbuf_try_pull(msg, sizeof *ps13);
|
||
if (!ps13) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_port_stats_from_ofp13(ps, ps13);
|
||
} else if (raw == OFPRAW_OFPST11_PORT_REPLY) {
|
||
const struct ofp11_port_stats *ps11;
|
||
|
||
ps11 = ofpbuf_try_pull(msg, sizeof *ps11);
|
||
if (!ps11) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_port_stats_from_ofp11(ps, ps11);
|
||
} else if (raw == OFPRAW_OFPST10_PORT_REPLY) {
|
||
const struct ofp10_port_stats *ps10;
|
||
|
||
ps10 = ofpbuf_try_pull(msg, sizeof *ps10);
|
||
if (!ps10) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_port_stats_from_ofp10(ps, ps10);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
bad_len:
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_PORT reply has %"PRIu32" leftover "
|
||
"bytes at end", ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
/* Parse a port status request message into a 16 bit OpenFlow 1.0
|
||
* port number and stores the latter in '*ofp10_port'.
|
||
* Returns 0 if successful, otherwise an OFPERR_* number. */
|
||
enum ofperr
|
||
ofputil_decode_port_stats_request(const struct ofp_header *request,
|
||
ofp_port_t *ofp10_port)
|
||
{
|
||
switch ((enum ofp_version)request->version) {
|
||
case OFP15_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP11_VERSION: {
|
||
const struct ofp11_port_stats_request *psr11 = ofpmsg_body(request);
|
||
return ofputil_port_from_ofp11(psr11->port_no, ofp10_port);
|
||
}
|
||
|
||
case OFP10_VERSION: {
|
||
const struct ofp10_port_stats_request *psr10 = ofpmsg_body(request);
|
||
*ofp10_port = u16_to_ofp(ntohs(psr10->port_no));
|
||
return 0;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Frees all of the "struct ofputil_bucket"s in the 'buckets' list. */
|
||
void
|
||
ofputil_bucket_list_destroy(struct list *buckets)
|
||
{
|
||
struct ofputil_bucket *bucket, *next_bucket;
|
||
|
||
LIST_FOR_EACH_SAFE (bucket, next_bucket, list_node, buckets) {
|
||
list_remove(&bucket->list_node);
|
||
free(bucket->ofpacts);
|
||
free(bucket);
|
||
}
|
||
}
|
||
|
||
/* Returns an OpenFlow group stats request for OpenFlow version 'ofp_version',
|
||
* that requests stats for group 'group_id'. (Use OFPG_ALL to request stats
|
||
* for all groups.)
|
||
*
|
||
* Group statistics include packet and byte counts for each group. */
|
||
struct ofpbuf *
|
||
ofputil_encode_group_stats_request(enum ofp_version ofp_version,
|
||
uint32_t group_id)
|
||
{
|
||
struct ofpbuf *request;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
ovs_fatal(0, "dump-group-stats needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp11_group_stats_request *req;
|
||
request = ofpraw_alloc(OFPRAW_OFPST11_GROUP_REQUEST, ofp_version, 0);
|
||
req = ofpbuf_put_zeros(request, sizeof *req);
|
||
req->group_id = htonl(group_id);
|
||
break;
|
||
}
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
/* Decodes the OpenFlow group description request in 'oh', returning the group
|
||
* whose description is requested, or OFPG_ALL if stats for all groups was
|
||
* requested. */
|
||
uint32_t
|
||
ofputil_decode_group_desc_request(const struct ofp_header *oh)
|
||
{
|
||
struct ofpbuf request;
|
||
enum ofpraw raw;
|
||
|
||
ofpbuf_use_const(&request, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&request);
|
||
if (raw == OFPRAW_OFPST11_GROUP_DESC_REQUEST) {
|
||
return OFPG_ALL;
|
||
} else if (raw == OFPRAW_OFPST15_GROUP_DESC_REQUEST) {
|
||
ovs_be32 *group_id = ofpbuf_pull(&request, sizeof *group_id);
|
||
return ntohl(*group_id);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Returns an OpenFlow group description request for OpenFlow version
|
||
* 'ofp_version', that requests stats for group 'group_id'. Use OFPG_ALL to
|
||
* request stats for all groups (OpenFlow 1.4 and earlier always request all
|
||
* groups).
|
||
*
|
||
* Group descriptions include the bucket and action configuration for each
|
||
* group. */
|
||
struct ofpbuf *
|
||
ofputil_encode_group_desc_request(enum ofp_version ofp_version,
|
||
uint32_t group_id)
|
||
{
|
||
struct ofpbuf *request;
|
||
ovs_be32 gid;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
ovs_fatal(0, "dump-groups needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST11_GROUP_DESC_REQUEST,
|
||
ofp_version, 0);
|
||
break;
|
||
case OFP15_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST15_GROUP_DESC_REQUEST,
|
||
ofp_version, 0);
|
||
gid = htonl(group_id);
|
||
ofpbuf_put(request, &gid, sizeof gid);
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
static void
|
||
ofputil_group_bucket_counters_to_ofp11(const struct ofputil_group_stats *gs,
|
||
struct ofp11_bucket_counter bucket_cnts[])
|
||
{
|
||
int i;
|
||
|
||
for (i = 0; i < gs->n_buckets; i++) {
|
||
bucket_cnts[i].packet_count = htonll(gs->bucket_stats[i].packet_count);
|
||
bucket_cnts[i].byte_count = htonll(gs->bucket_stats[i].byte_count);
|
||
}
|
||
}
|
||
|
||
static void
|
||
ofputil_group_stats_to_ofp11(const struct ofputil_group_stats *gs,
|
||
struct ofp11_group_stats *gs11, size_t length,
|
||
struct ofp11_bucket_counter bucket_cnts[])
|
||
{
|
||
memset(gs11, 0, sizeof *gs11);
|
||
gs11->length = htons(length);
|
||
gs11->group_id = htonl(gs->group_id);
|
||
gs11->ref_count = htonl(gs->ref_count);
|
||
gs11->packet_count = htonll(gs->packet_count);
|
||
gs11->byte_count = htonll(gs->byte_count);
|
||
ofputil_group_bucket_counters_to_ofp11(gs, bucket_cnts);
|
||
}
|
||
|
||
static void
|
||
ofputil_group_stats_to_ofp13(const struct ofputil_group_stats *gs,
|
||
struct ofp13_group_stats *gs13, size_t length,
|
||
struct ofp11_bucket_counter bucket_cnts[])
|
||
{
|
||
ofputil_group_stats_to_ofp11(gs, &gs13->gs, length, bucket_cnts);
|
||
gs13->duration_sec = htonl(gs->duration_sec);
|
||
gs13->duration_nsec = htonl(gs->duration_nsec);
|
||
|
||
}
|
||
|
||
/* Encodes 'gs' properly for the format of the list of group statistics
|
||
* replies already begun in 'replies' and appends it to the list. 'replies'
|
||
* must have originally been initialized with ofpmp_init(). */
|
||
void
|
||
ofputil_append_group_stats(struct list *replies,
|
||
const struct ofputil_group_stats *gs)
|
||
{
|
||
size_t bucket_counter_size;
|
||
struct ofp11_bucket_counter *bucket_counters;
|
||
size_t length;
|
||
|
||
bucket_counter_size = gs->n_buckets * sizeof(struct ofp11_bucket_counter);
|
||
|
||
switch (ofpmp_version(replies)) {
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:{
|
||
struct ofp11_group_stats *gs11;
|
||
|
||
length = sizeof *gs11 + bucket_counter_size;
|
||
gs11 = ofpmp_append(replies, length);
|
||
bucket_counters = (struct ofp11_bucket_counter *)(gs11 + 1);
|
||
ofputil_group_stats_to_ofp11(gs, gs11, length, bucket_counters);
|
||
break;
|
||
}
|
||
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp13_group_stats *gs13;
|
||
|
||
length = sizeof *gs13 + bucket_counter_size;
|
||
gs13 = ofpmp_append(replies, length);
|
||
bucket_counters = (struct ofp11_bucket_counter *)(gs13 + 1);
|
||
ofputil_group_stats_to_ofp13(gs, gs13, length, bucket_counters);
|
||
break;
|
||
}
|
||
|
||
case OFP10_VERSION:
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
/* Returns an OpenFlow group features request for OpenFlow version
|
||
* 'ofp_version'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_group_features_request(enum ofp_version ofp_version)
|
||
{
|
||
struct ofpbuf *request = NULL;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION:
|
||
case OFP11_VERSION:
|
||
ovs_fatal(0, "dump-group-features needs OpenFlow 1.2 or later "
|
||
"(\'-O OpenFlow12\')");
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
request = ofpraw_alloc(OFPRAW_OFPST12_GROUP_FEATURES_REQUEST,
|
||
ofp_version, 0);
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
/* Returns a OpenFlow message that encodes 'features' properly as a reply to
|
||
* group features request 'request'. */
|
||
struct ofpbuf *
|
||
ofputil_encode_group_features_reply(
|
||
const struct ofputil_group_features *features,
|
||
const struct ofp_header *request)
|
||
{
|
||
struct ofp12_group_features_stats *ogf;
|
||
struct ofpbuf *reply;
|
||
int i;
|
||
|
||
reply = ofpraw_alloc_xid(OFPRAW_OFPST12_GROUP_FEATURES_REPLY,
|
||
request->version, request->xid, 0);
|
||
ogf = ofpbuf_put_zeros(reply, sizeof *ogf);
|
||
ogf->types = htonl(features->types);
|
||
ogf->capabilities = htonl(features->capabilities);
|
||
for (i = 0; i < OFPGT12_N_TYPES; i++) {
|
||
ogf->max_groups[i] = htonl(features->max_groups[i]);
|
||
ogf->actions[i] = ofpact_bitmap_to_openflow(features->ofpacts[i],
|
||
request->version);
|
||
}
|
||
|
||
return reply;
|
||
}
|
||
|
||
/* Decodes group features reply 'oh' into 'features'. */
|
||
void
|
||
ofputil_decode_group_features_reply(const struct ofp_header *oh,
|
||
struct ofputil_group_features *features)
|
||
{
|
||
const struct ofp12_group_features_stats *ogf = ofpmsg_body(oh);
|
||
int i;
|
||
|
||
features->types = ntohl(ogf->types);
|
||
features->capabilities = ntohl(ogf->capabilities);
|
||
for (i = 0; i < OFPGT12_N_TYPES; i++) {
|
||
features->max_groups[i] = ntohl(ogf->max_groups[i]);
|
||
features->ofpacts[i] = ofpact_bitmap_from_openflow(
|
||
ogf->actions[i], oh->version);
|
||
}
|
||
}
|
||
|
||
/* Parse a group status request message into a 32 bit OpenFlow 1.1
|
||
* group ID and stores the latter in '*group_id'.
|
||
* Returns 0 if successful, otherwise an OFPERR_* number. */
|
||
enum ofperr
|
||
ofputil_decode_group_stats_request(const struct ofp_header *request,
|
||
uint32_t *group_id)
|
||
{
|
||
const struct ofp11_group_stats_request *gsr11 = ofpmsg_body(request);
|
||
*group_id = ntohl(gsr11->group_id);
|
||
return 0;
|
||
}
|
||
|
||
/* Converts a group stats reply in 'msg' into an abstract ofputil_group_stats
|
||
* in 'gs'. Assigns freshly allocated memory to gs->bucket_stats for the
|
||
* caller to eventually free.
|
||
*
|
||
* Multiple group stats replies can be packed into a single OpenFlow message.
|
||
* Calling this function multiple times for a single 'msg' iterates through the
|
||
* replies. The caller must initially leave 'msg''s layer pointers null and
|
||
* not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_group_stats_reply(struct ofpbuf *msg,
|
||
struct ofputil_group_stats *gs)
|
||
{
|
||
struct ofp11_bucket_counter *obc;
|
||
struct ofp11_group_stats *ogs11;
|
||
enum ofpraw raw;
|
||
enum ofperr error;
|
||
size_t base_len;
|
||
size_t length;
|
||
size_t i;
|
||
|
||
gs->bucket_stats = NULL;
|
||
error = (msg->frame
|
||
? ofpraw_decode(&raw, msg->frame)
|
||
: ofpraw_pull(&raw, msg));
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
if (raw == OFPRAW_OFPST11_GROUP_REPLY) {
|
||
base_len = sizeof *ogs11;
|
||
ogs11 = ofpbuf_try_pull(msg, sizeof *ogs11);
|
||
gs->duration_sec = gs->duration_nsec = UINT32_MAX;
|
||
} else if (raw == OFPRAW_OFPST13_GROUP_REPLY) {
|
||
struct ofp13_group_stats *ogs13;
|
||
|
||
base_len = sizeof *ogs13;
|
||
ogs13 = ofpbuf_try_pull(msg, sizeof *ogs13);
|
||
if (ogs13) {
|
||
ogs11 = &ogs13->gs;
|
||
gs->duration_sec = ntohl(ogs13->duration_sec);
|
||
gs->duration_nsec = ntohl(ogs13->duration_nsec);
|
||
} else {
|
||
ogs11 = NULL;
|
||
}
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
if (!ogs11) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "%s reply has %"PRIu32" leftover bytes at end",
|
||
ofpraw_get_name(raw), ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
length = ntohs(ogs11->length);
|
||
if (length < sizeof base_len) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "%s reply claims invalid length %"PRIuSIZE,
|
||
ofpraw_get_name(raw), length);
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
gs->group_id = ntohl(ogs11->group_id);
|
||
gs->ref_count = ntohl(ogs11->ref_count);
|
||
gs->packet_count = ntohll(ogs11->packet_count);
|
||
gs->byte_count = ntohll(ogs11->byte_count);
|
||
|
||
gs->n_buckets = (length - base_len) / sizeof *obc;
|
||
obc = ofpbuf_try_pull(msg, gs->n_buckets * sizeof *obc);
|
||
if (!obc) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "%s reply has %"PRIu32" leftover bytes at end",
|
||
ofpraw_get_name(raw), ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
gs->bucket_stats = xmalloc(gs->n_buckets * sizeof *gs->bucket_stats);
|
||
for (i = 0; i < gs->n_buckets; i++) {
|
||
gs->bucket_stats[i].packet_count = ntohll(obc[i].packet_count);
|
||
gs->bucket_stats[i].byte_count = ntohll(obc[i].byte_count);
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Appends a group stats reply that contains the data in 'gds' to those already
|
||
* present in the list of ofpbufs in 'replies'. 'replies' should have been
|
||
* initialized with ofpmp_init(). */
|
||
void
|
||
ofputil_append_group_desc_reply(const struct ofputil_group_desc *gds,
|
||
struct list *buckets,
|
||
struct list *replies)
|
||
{
|
||
struct ofpbuf *reply = ofpbuf_from_list(list_back(replies));
|
||
enum ofp_version version = ofpmp_version(replies);
|
||
struct ofp11_group_desc_stats *ogds;
|
||
struct ofputil_bucket *bucket;
|
||
size_t start_ogds;
|
||
|
||
start_ogds = ofpbuf_size(reply);
|
||
ofpbuf_put_zeros(reply, sizeof *ogds);
|
||
LIST_FOR_EACH (bucket, list_node, buckets) {
|
||
struct ofp11_bucket *ob;
|
||
size_t start_ob;
|
||
|
||
start_ob = ofpbuf_size(reply);
|
||
ofpbuf_put_zeros(reply, sizeof *ob);
|
||
ofpacts_put_openflow_actions(bucket->ofpacts, bucket->ofpacts_len,
|
||
reply, version);
|
||
ob = ofpbuf_at_assert(reply, start_ob, sizeof *ob);
|
||
ob->len = htons(ofpbuf_size(reply) - start_ob);
|
||
ob->weight = htons(bucket->weight);
|
||
ob->watch_port = ofputil_port_to_ofp11(bucket->watch_port);
|
||
ob->watch_group = htonl(bucket->watch_group);
|
||
}
|
||
ogds = ofpbuf_at_assert(reply, start_ogds, sizeof *ogds);
|
||
ogds->length = htons(ofpbuf_size(reply) - start_ogds);
|
||
ogds->type = gds->type;
|
||
ogds->group_id = htonl(gds->group_id);
|
||
|
||
ofpmp_postappend(replies, start_ogds);
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_buckets(struct ofpbuf *msg, size_t buckets_length,
|
||
enum ofp_version version, struct list *buckets)
|
||
{
|
||
struct ofp11_bucket *ob;
|
||
|
||
list_init(buckets);
|
||
while (buckets_length > 0) {
|
||
struct ofputil_bucket *bucket;
|
||
struct ofpbuf ofpacts;
|
||
enum ofperr error;
|
||
size_t ob_len;
|
||
|
||
ob = (buckets_length >= sizeof *ob
|
||
? ofpbuf_try_pull(msg, sizeof *ob)
|
||
: NULL);
|
||
if (!ob) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "buckets end with %"PRIuSIZE" leftover bytes",
|
||
buckets_length);
|
||
return OFPERR_OFPGMFC_BAD_BUCKET;
|
||
}
|
||
|
||
ob_len = ntohs(ob->len);
|
||
if (ob_len < sizeof *ob) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OpenFlow message bucket length "
|
||
"%"PRIuSIZE" is not valid", ob_len);
|
||
return OFPERR_OFPGMFC_BAD_BUCKET;
|
||
} else if (ob_len > buckets_length) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OpenFlow message bucket length "
|
||
"%"PRIuSIZE" exceeds remaining buckets data size %"PRIuSIZE,
|
||
ob_len, buckets_length);
|
||
return OFPERR_OFPGMFC_BAD_BUCKET;
|
||
}
|
||
buckets_length -= ob_len;
|
||
|
||
ofpbuf_init(&ofpacts, 0);
|
||
error = ofpacts_pull_openflow_actions(msg, ob_len - sizeof *ob,
|
||
version, &ofpacts);
|
||
if (error) {
|
||
ofpbuf_uninit(&ofpacts);
|
||
ofputil_bucket_list_destroy(buckets);
|
||
return error;
|
||
}
|
||
|
||
bucket = xzalloc(sizeof *bucket);
|
||
bucket->weight = ntohs(ob->weight);
|
||
error = ofputil_port_from_ofp11(ob->watch_port, &bucket->watch_port);
|
||
if (error) {
|
||
ofpbuf_uninit(&ofpacts);
|
||
ofputil_bucket_list_destroy(buckets);
|
||
return OFPERR_OFPGMFC_BAD_WATCH;
|
||
}
|
||
bucket->watch_group = ntohl(ob->watch_group);
|
||
bucket->ofpacts = ofpbuf_steal_data(&ofpacts);
|
||
bucket->ofpacts_len = ofpbuf_size(&ofpacts);
|
||
list_push_back(buckets, &bucket->list_node);
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Converts a group description reply in 'msg' into an abstract
|
||
* ofputil_group_desc in 'gd'.
|
||
*
|
||
* Multiple group description replies can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the replies. The caller must initially leave 'msg''s layer pointers
|
||
* null and not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_group_desc_reply(struct ofputil_group_desc *gd,
|
||
struct ofpbuf *msg, enum ofp_version version)
|
||
{
|
||
struct ofp11_group_desc_stats *ogds;
|
||
size_t length;
|
||
|
||
if (!msg->frame) {
|
||
ofpraw_pull_assert(msg);
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
}
|
||
|
||
ogds = ofpbuf_try_pull(msg, sizeof *ogds);
|
||
if (!ogds) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST11_GROUP_DESC reply has %"PRIu32" "
|
||
"leftover bytes at end", ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
gd->type = ogds->type;
|
||
gd->group_id = ntohl(ogds->group_id);
|
||
|
||
length = ntohs(ogds->length);
|
||
if (length < sizeof *ogds || length - sizeof *ogds > ofpbuf_size(msg)) {
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST11_GROUP_DESC reply claims invalid "
|
||
"length %"PRIuSIZE, length);
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
return ofputil_pull_buckets(msg, length - sizeof *ogds, version,
|
||
&gd->buckets);
|
||
}
|
||
|
||
/* Converts abstract group mod 'gm' into a message for OpenFlow version
|
||
* 'ofp_version' and returns the message. */
|
||
struct ofpbuf *
|
||
ofputil_encode_group_mod(enum ofp_version ofp_version,
|
||
const struct ofputil_group_mod *gm)
|
||
{
|
||
struct ofpbuf *b;
|
||
struct ofp11_group_mod *ogm;
|
||
size_t start_ogm;
|
||
size_t start_bucket;
|
||
struct ofputil_bucket *bucket;
|
||
struct ofp11_bucket *ob;
|
||
|
||
switch (ofp_version) {
|
||
case OFP10_VERSION: {
|
||
if (gm->command == OFPGC11_ADD) {
|
||
ovs_fatal(0, "add-group needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
} else if (gm->command == OFPGC11_MODIFY) {
|
||
ovs_fatal(0, "mod-group needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
} else {
|
||
ovs_fatal(0, "del-groups needs OpenFlow 1.1 or later "
|
||
"(\'-O OpenFlow11\')");
|
||
}
|
||
}
|
||
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION:
|
||
b = ofpraw_alloc(OFPRAW_OFPT11_GROUP_MOD, ofp_version, 0);
|
||
start_ogm = ofpbuf_size(b);
|
||
ofpbuf_put_zeros(b, sizeof *ogm);
|
||
|
||
LIST_FOR_EACH (bucket, list_node, &gm->buckets) {
|
||
start_bucket = ofpbuf_size(b);
|
||
ofpbuf_put_zeros(b, sizeof *ob);
|
||
if (bucket->ofpacts && bucket->ofpacts_len) {
|
||
ofpacts_put_openflow_actions(bucket->ofpacts,
|
||
bucket->ofpacts_len, b,
|
||
ofp_version);
|
||
}
|
||
ob = ofpbuf_at_assert(b, start_bucket, sizeof *ob);
|
||
ob->len = htons(ofpbuf_size(b) - start_bucket);;
|
||
ob->weight = htons(bucket->weight);
|
||
ob->watch_port = ofputil_port_to_ofp11(bucket->watch_port);
|
||
ob->watch_group = htonl(bucket->watch_group);
|
||
}
|
||
ogm = ofpbuf_at_assert(b, start_ogm, sizeof *ogm);
|
||
ogm->command = htons(gm->command);
|
||
ogm->type = gm->type;
|
||
ogm->group_id = htonl(gm->group_id);
|
||
|
||
break;
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return b;
|
||
}
|
||
|
||
/* Converts OpenFlow group mod message 'oh' into an abstract group mod in
|
||
* 'gm'. Returns 0 if successful, otherwise an OpenFlow error code. */
|
||
enum ofperr
|
||
ofputil_decode_group_mod(const struct ofp_header *oh,
|
||
struct ofputil_group_mod *gm)
|
||
{
|
||
const struct ofp11_group_mod *ogm;
|
||
struct ofpbuf msg;
|
||
struct ofputil_bucket *bucket;
|
||
enum ofperr err;
|
||
|
||
ofpbuf_use_const(&msg, oh, ntohs(oh->length));
|
||
ofpraw_pull_assert(&msg);
|
||
|
||
ogm = ofpbuf_pull(&msg, sizeof *ogm);
|
||
gm->command = ntohs(ogm->command);
|
||
gm->type = ogm->type;
|
||
gm->group_id = ntohl(ogm->group_id);
|
||
|
||
err = ofputil_pull_buckets(&msg, ofpbuf_size(&msg), oh->version, &gm->buckets);
|
||
if (err) {
|
||
return err;
|
||
}
|
||
|
||
LIST_FOR_EACH (bucket, list_node, &gm->buckets) {
|
||
switch (gm->type) {
|
||
case OFPGT11_ALL:
|
||
case OFPGT11_INDIRECT:
|
||
if (ofputil_bucket_has_liveness(bucket)) {
|
||
return OFPERR_OFPGMFC_WATCH_UNSUPPORTED;
|
||
}
|
||
break;
|
||
case OFPGT11_SELECT:
|
||
break;
|
||
case OFPGT11_FF:
|
||
if (!ofputil_bucket_has_liveness(bucket)) {
|
||
return OFPERR_OFPGMFC_INVALID_GROUP;
|
||
}
|
||
break;
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
/* Parse a queue status request message into 'oqsr'.
|
||
* Returns 0 if successful, otherwise an OFPERR_* number. */
|
||
enum ofperr
|
||
ofputil_decode_queue_stats_request(const struct ofp_header *request,
|
||
struct ofputil_queue_stats_request *oqsr)
|
||
{
|
||
switch ((enum ofp_version)request->version) {
|
||
case OFP15_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP11_VERSION: {
|
||
const struct ofp11_queue_stats_request *qsr11 = ofpmsg_body(request);
|
||
oqsr->queue_id = ntohl(qsr11->queue_id);
|
||
return ofputil_port_from_ofp11(qsr11->port_no, &oqsr->port_no);
|
||
}
|
||
|
||
case OFP10_VERSION: {
|
||
const struct ofp10_queue_stats_request *qsr10 = ofpmsg_body(request);
|
||
oqsr->queue_id = ntohl(qsr10->queue_id);
|
||
oqsr->port_no = u16_to_ofp(ntohs(qsr10->port_no));
|
||
/* OF 1.0 uses OFPP_ALL for OFPP_ANY */
|
||
if (oqsr->port_no == OFPP_ALL) {
|
||
oqsr->port_no = OFPP_ANY;
|
||
}
|
||
return 0;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
/* Encode a queue statsrequest for 'oqsr', the encoded message
|
||
* will be fore Open Flow version 'ofp_version'. Returns message
|
||
* as a struct ofpbuf. Returns encoded message on success, NULL on error */
|
||
struct ofpbuf *
|
||
ofputil_encode_queue_stats_request(enum ofp_version ofp_version,
|
||
const struct ofputil_queue_stats_request *oqsr)
|
||
{
|
||
struct ofpbuf *request;
|
||
|
||
switch (ofp_version) {
|
||
case OFP11_VERSION:
|
||
case OFP12_VERSION:
|
||
case OFP13_VERSION:
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp11_queue_stats_request *req;
|
||
request = ofpraw_alloc(OFPRAW_OFPST11_QUEUE_REQUEST, ofp_version, 0);
|
||
req = ofpbuf_put_zeros(request, sizeof *req);
|
||
req->port_no = ofputil_port_to_ofp11(oqsr->port_no);
|
||
req->queue_id = htonl(oqsr->queue_id);
|
||
break;
|
||
}
|
||
case OFP10_VERSION: {
|
||
struct ofp10_queue_stats_request *req;
|
||
request = ofpraw_alloc(OFPRAW_OFPST10_QUEUE_REQUEST, ofp_version, 0);
|
||
req = ofpbuf_put_zeros(request, sizeof *req);
|
||
/* OpenFlow 1.0 needs OFPP_ALL instead of OFPP_ANY */
|
||
req->port_no = htons(ofp_to_u16(oqsr->port_no == OFPP_ANY
|
||
? OFPP_ALL : oqsr->port_no));
|
||
req->queue_id = htonl(oqsr->queue_id);
|
||
break;
|
||
}
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
return request;
|
||
}
|
||
|
||
/* Returns the number of queue stats elements in OFPTYPE_QUEUE_STATS_REPLY
|
||
* message 'oh'. */
|
||
size_t
|
||
ofputil_count_queue_stats(const struct ofp_header *oh)
|
||
{
|
||
struct ofputil_queue_stats qs;
|
||
struct ofpbuf b;
|
||
size_t n = 0;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
ofpraw_pull_assert(&b);
|
||
while (!ofputil_decode_queue_stats(&qs, &b)) {
|
||
n++;
|
||
}
|
||
return n;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_queue_stats_from_ofp10(struct ofputil_queue_stats *oqs,
|
||
const struct ofp10_queue_stats *qs10)
|
||
{
|
||
oqs->port_no = u16_to_ofp(ntohs(qs10->port_no));
|
||
oqs->queue_id = ntohl(qs10->queue_id);
|
||
oqs->tx_bytes = ntohll(get_32aligned_be64(&qs10->tx_bytes));
|
||
oqs->tx_packets = ntohll(get_32aligned_be64(&qs10->tx_packets));
|
||
oqs->tx_errors = ntohll(get_32aligned_be64(&qs10->tx_errors));
|
||
oqs->duration_sec = oqs->duration_nsec = UINT32_MAX;
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_queue_stats_from_ofp11(struct ofputil_queue_stats *oqs,
|
||
const struct ofp11_queue_stats *qs11)
|
||
{
|
||
enum ofperr error;
|
||
|
||
error = ofputil_port_from_ofp11(qs11->port_no, &oqs->port_no);
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
oqs->queue_id = ntohl(qs11->queue_id);
|
||
oqs->tx_bytes = ntohll(qs11->tx_bytes);
|
||
oqs->tx_packets = ntohll(qs11->tx_packets);
|
||
oqs->tx_errors = ntohll(qs11->tx_errors);
|
||
oqs->duration_sec = oqs->duration_nsec = UINT32_MAX;
|
||
|
||
return 0;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_queue_stats_from_ofp13(struct ofputil_queue_stats *oqs,
|
||
const struct ofp13_queue_stats *qs13)
|
||
{
|
||
enum ofperr error = ofputil_queue_stats_from_ofp11(oqs, &qs13->qs);
|
||
if (!error) {
|
||
oqs->duration_sec = ntohl(qs13->duration_sec);
|
||
oqs->duration_nsec = ntohl(qs13->duration_nsec);
|
||
}
|
||
|
||
return error;
|
||
}
|
||
|
||
static enum ofperr
|
||
ofputil_pull_ofp14_queue_stats(struct ofputil_queue_stats *oqs,
|
||
struct ofpbuf *msg)
|
||
{
|
||
const struct ofp14_queue_stats *qs14;
|
||
size_t len;
|
||
|
||
qs14 = ofpbuf_try_pull(msg, sizeof *qs14);
|
||
if (!qs14) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
len = ntohs(qs14->length);
|
||
if (len < sizeof *qs14 || len - sizeof *qs14 > ofpbuf_size(msg)) {
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
ofpbuf_pull(msg, len - sizeof *qs14);
|
||
|
||
/* No properties yet defined, so ignore them for now. */
|
||
|
||
return ofputil_queue_stats_from_ofp13(oqs, &qs14->qs);
|
||
}
|
||
|
||
/* Converts an OFPST_QUEUE_STATS reply in 'msg' into an abstract
|
||
* ofputil_queue_stats in 'qs'.
|
||
*
|
||
* Multiple OFPST_QUEUE_STATS replies can be packed into a single OpenFlow
|
||
* message. Calling this function multiple times for a single 'msg' iterates
|
||
* through the replies. The caller must initially leave 'msg''s layer pointers
|
||
* null and not modify them between calls.
|
||
*
|
||
* Returns 0 if successful, EOF if no replies were left in this 'msg',
|
||
* otherwise a positive errno value. */
|
||
int
|
||
ofputil_decode_queue_stats(struct ofputil_queue_stats *qs, struct ofpbuf *msg)
|
||
{
|
||
enum ofperr error;
|
||
enum ofpraw raw;
|
||
|
||
error = (msg->frame
|
||
? ofpraw_decode(&raw, msg->frame)
|
||
: ofpraw_pull(&raw, msg));
|
||
if (error) {
|
||
return error;
|
||
}
|
||
|
||
if (!ofpbuf_size(msg)) {
|
||
return EOF;
|
||
} else if (raw == OFPRAW_OFPST14_QUEUE_REPLY) {
|
||
return ofputil_pull_ofp14_queue_stats(qs, msg);
|
||
} else if (raw == OFPRAW_OFPST13_QUEUE_REPLY) {
|
||
const struct ofp13_queue_stats *qs13;
|
||
|
||
qs13 = ofpbuf_try_pull(msg, sizeof *qs13);
|
||
if (!qs13) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_queue_stats_from_ofp13(qs, qs13);
|
||
} else if (raw == OFPRAW_OFPST11_QUEUE_REPLY) {
|
||
const struct ofp11_queue_stats *qs11;
|
||
|
||
qs11 = ofpbuf_try_pull(msg, sizeof *qs11);
|
||
if (!qs11) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_queue_stats_from_ofp11(qs, qs11);
|
||
} else if (raw == OFPRAW_OFPST10_QUEUE_REPLY) {
|
||
const struct ofp10_queue_stats *qs10;
|
||
|
||
qs10 = ofpbuf_try_pull(msg, sizeof *qs10);
|
||
if (!qs10) {
|
||
goto bad_len;
|
||
}
|
||
return ofputil_queue_stats_from_ofp10(qs, qs10);
|
||
} else {
|
||
OVS_NOT_REACHED();
|
||
}
|
||
|
||
bad_len:
|
||
VLOG_WARN_RL(&bad_ofmsg_rl, "OFPST_QUEUE reply has %"PRIu32" leftover "
|
||
"bytes at end", ofpbuf_size(msg));
|
||
return OFPERR_OFPBRC_BAD_LEN;
|
||
}
|
||
|
||
static void
|
||
ofputil_queue_stats_to_ofp10(const struct ofputil_queue_stats *oqs,
|
||
struct ofp10_queue_stats *qs10)
|
||
{
|
||
qs10->port_no = htons(ofp_to_u16(oqs->port_no));
|
||
memset(qs10->pad, 0, sizeof qs10->pad);
|
||
qs10->queue_id = htonl(oqs->queue_id);
|
||
put_32aligned_be64(&qs10->tx_bytes, htonll(oqs->tx_bytes));
|
||
put_32aligned_be64(&qs10->tx_packets, htonll(oqs->tx_packets));
|
||
put_32aligned_be64(&qs10->tx_errors, htonll(oqs->tx_errors));
|
||
}
|
||
|
||
static void
|
||
ofputil_queue_stats_to_ofp11(const struct ofputil_queue_stats *oqs,
|
||
struct ofp11_queue_stats *qs11)
|
||
{
|
||
qs11->port_no = ofputil_port_to_ofp11(oqs->port_no);
|
||
qs11->queue_id = htonl(oqs->queue_id);
|
||
qs11->tx_bytes = htonll(oqs->tx_bytes);
|
||
qs11->tx_packets = htonll(oqs->tx_packets);
|
||
qs11->tx_errors = htonll(oqs->tx_errors);
|
||
}
|
||
|
||
static void
|
||
ofputil_queue_stats_to_ofp13(const struct ofputil_queue_stats *oqs,
|
||
struct ofp13_queue_stats *qs13)
|
||
{
|
||
ofputil_queue_stats_to_ofp11(oqs, &qs13->qs);
|
||
if (oqs->duration_sec != UINT32_MAX) {
|
||
qs13->duration_sec = htonl(oqs->duration_sec);
|
||
qs13->duration_nsec = htonl(oqs->duration_nsec);
|
||
} else {
|
||
qs13->duration_sec = OVS_BE32_MAX;
|
||
qs13->duration_nsec = OVS_BE32_MAX;
|
||
}
|
||
}
|
||
|
||
static void
|
||
ofputil_queue_stats_to_ofp14(const struct ofputil_queue_stats *oqs,
|
||
struct ofp14_queue_stats *qs14)
|
||
{
|
||
qs14->length = htons(sizeof *qs14);
|
||
memset(qs14->pad, 0, sizeof qs14->pad);
|
||
ofputil_queue_stats_to_ofp13(oqs, &qs14->qs);
|
||
}
|
||
|
||
|
||
/* Encode a queue stat for 'oqs' and append it to 'replies'. */
|
||
void
|
||
ofputil_append_queue_stat(struct list *replies,
|
||
const struct ofputil_queue_stats *oqs)
|
||
{
|
||
switch (ofpmp_version(replies)) {
|
||
case OFP13_VERSION: {
|
||
struct ofp13_queue_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_queue_stats_to_ofp13(oqs, reply);
|
||
break;
|
||
}
|
||
|
||
case OFP12_VERSION:
|
||
case OFP11_VERSION: {
|
||
struct ofp11_queue_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_queue_stats_to_ofp11(oqs, reply);
|
||
break;
|
||
}
|
||
|
||
case OFP10_VERSION: {
|
||
struct ofp10_queue_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_queue_stats_to_ofp10(oqs, reply);
|
||
break;
|
||
}
|
||
|
||
case OFP14_VERSION:
|
||
case OFP15_VERSION: {
|
||
struct ofp14_queue_stats *reply = ofpmp_append(replies, sizeof *reply);
|
||
ofputil_queue_stats_to_ofp14(oqs, reply);
|
||
break;
|
||
}
|
||
|
||
default:
|
||
OVS_NOT_REACHED();
|
||
}
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_bundle_ctrl(const struct ofp_header *oh,
|
||
struct ofputil_bundle_ctrl_msg *msg)
|
||
{
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
const struct ofp14_bundle_ctrl_msg *m;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
ovs_assert(raw == OFPRAW_OFPT14_BUNDLE_CONTROL);
|
||
|
||
m = ofpbuf_l3(&b);
|
||
msg->bundle_id = ntohl(m->bundle_id);
|
||
msg->type = ntohs(m->type);
|
||
msg->flags = ntohs(m->flags);
|
||
|
||
return 0;
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_bundle_ctrl_reply(const struct ofp_header *oh,
|
||
struct ofputil_bundle_ctrl_msg *msg)
|
||
{
|
||
struct ofpbuf *buf;
|
||
struct ofp14_bundle_ctrl_msg *m;
|
||
|
||
buf = ofpraw_alloc_reply(OFPRAW_OFPT14_BUNDLE_CONTROL, oh, 0);
|
||
m = ofpbuf_put_zeros(buf, sizeof *m);
|
||
|
||
m->bundle_id = htonl(msg->bundle_id);
|
||
m->type = htons(msg->type);
|
||
m->flags = htons(msg->flags);
|
||
|
||
return buf;
|
||
}
|
||
|
||
enum ofperr
|
||
ofputil_decode_bundle_add(const struct ofp_header *oh,
|
||
struct ofputil_bundle_add_msg *msg)
|
||
{
|
||
const struct ofp14_bundle_ctrl_msg *m;
|
||
struct ofpbuf b;
|
||
enum ofpraw raw;
|
||
size_t inner_len;
|
||
|
||
ofpbuf_use_const(&b, oh, ntohs(oh->length));
|
||
raw = ofpraw_pull_assert(&b);
|
||
ovs_assert(raw == OFPRAW_OFPT14_BUNDLE_ADD_MESSAGE);
|
||
|
||
m = ofpbuf_pull(&b, sizeof *m);
|
||
msg->bundle_id = ntohl(m->bundle_id);
|
||
msg->flags = ntohs(m->flags);
|
||
|
||
msg->msg = ofpbuf_data(&b);
|
||
inner_len = ntohs(msg->msg->length);
|
||
if (inner_len < sizeof(struct ofp_header) || inner_len > ofpbuf_size(&b)) {
|
||
return OFPERR_OFPBFC_MSG_BAD_LEN;
|
||
}
|
||
|
||
return 0;
|
||
}
|
||
|
||
struct ofpbuf *
|
||
ofputil_encode_bundle_add(enum ofp_version ofp_version,
|
||
struct ofputil_bundle_add_msg *msg)
|
||
{
|
||
struct ofpbuf *request;
|
||
struct ofp14_bundle_ctrl_msg *m;
|
||
|
||
request = ofpraw_alloc(OFPRAW_OFPT14_BUNDLE_ADD_MESSAGE, ofp_version, 0);
|
||
m = ofpbuf_put_zeros(request, sizeof *m);
|
||
|
||
m->bundle_id = htonl(msg->bundle_id);
|
||
m->flags = htons(msg->flags);
|
||
ofpbuf_put(request, msg->msg, ntohs(msg->msg->length));
|
||
|
||
return request;
|
||
}
|