@@ -214,7 +214,7 @@ static VirtIOFeature feature_sizes[] = {
.end = endof(struct virtio_net_config, mtu)},
{.flags = 1ULL << VIRTIO_NET_F_SPEED_DUPLEX,
.end = endof(struct virtio_net_config, duplex)},
- {.flags = 1ULL << VIRTIO_NET_F_RSS,
+ {.flags = (1ULL << VIRTIO_NET_F_RSS) | (1ULL << VIRTIO_NET_F_HASH_REPORT),
.end = endof(struct virtio_net_config, supported_hash_types)},
{}
};
@@ -248,7 +248,8 @@ static void virtio_net_get_config(VirtIODevice *vdev, uint8_t *config)
netcfg.duplex = n->net_conf.duplex;
netcfg.rss_max_key_size = VIRTIO_NET_RSS_MAX_KEY_SIZE;
virtio_stw_p(vdev, &netcfg.rss_max_indirection_table_length,
- VIRTIO_NET_RSS_MAX_TABLE_LEN);
+ virtio_host_has_feature(vdev, VIRTIO_NET_F_RSS) ?
+ VIRTIO_NET_RSS_MAX_TABLE_LEN : 1);
virtio_stl_p(vdev, &netcfg.supported_hash_types,
VIRTIO_NET_RSS_SUPPORTED_HASHES);
memcpy(config, &netcfg, n->config_size);
@@ -674,7 +675,7 @@ static int peer_has_ufo(VirtIONet *n)
}
static void virtio_net_set_mrg_rx_bufs(VirtIONet *n, int mergeable_rx_bufs,
- int version_1)
+ int version_1, int hash_report)
{
int i;
NetClientState *nc;
@@ -682,7 +683,10 @@ static void virtio_net_set_mrg_rx_bufs(VirtIONet *n, int mergeable_rx_bufs,
n->mergeable_rx_bufs = mergeable_rx_bufs;
if (version_1) {
- n->guest_hdr_len = sizeof(struct virtio_net_hdr_mrg_rxbuf);
+ n->guest_hdr_len = hash_report ?
+ sizeof(struct virtio_net_hdr_v1_hash) :
+ sizeof(struct virtio_net_hdr_mrg_rxbuf);
+ n->rss_data.populate_hash = true;
} else {
n->guest_hdr_len = n->mergeable_rx_bufs ?
sizeof(struct virtio_net_hdr_mrg_rxbuf) :
@@ -803,6 +807,8 @@ static uint64_t virtio_net_get_features(VirtIODevice *vdev, uint64_t features,
virtio_clear_feature(&features, VIRTIO_NET_F_GUEST_TSO4);
virtio_clear_feature(&features, VIRTIO_NET_F_GUEST_TSO6);
virtio_clear_feature(&features, VIRTIO_NET_F_GUEST_ECN);
+
+ virtio_clear_feature(&features, VIRTIO_NET_F_HASH_REPORT);
}
if (!peer_has_vnet_hdr(n) || !peer_has_ufo(n)) {
@@ -815,6 +821,7 @@ static uint64_t virtio_net_get_features(VirtIODevice *vdev, uint64_t features,
}
virtio_clear_feature(&features, VIRTIO_NET_F_RSS);
+ virtio_clear_feature(&features, VIRTIO_NET_F_HASH_REPORT);
features = vhost_net_get_features(get_vhost_net(nc->peer), features);
vdev->backend_features = features;
@@ -981,12 +988,15 @@ static void virtio_net_set_features(VirtIODevice *vdev, uint64_t features)
virtio_has_feature(features,
VIRTIO_NET_F_MRG_RXBUF),
virtio_has_feature(features,
- VIRTIO_F_VERSION_1));
+ VIRTIO_F_VERSION_1),
+ virtio_has_feature(features,
+ VIRTIO_NET_F_HASH_REPORT));
n->rsc4_enabled = virtio_has_feature(features, VIRTIO_NET_F_RSC_EXT) &&
virtio_has_feature(features, VIRTIO_NET_F_GUEST_TSO4);
n->rsc6_enabled = virtio_has_feature(features, VIRTIO_NET_F_RSC_EXT) &&
virtio_has_feature(features, VIRTIO_NET_F_GUEST_TSO6);
+ n->rss_data.redirect = virtio_has_feature(features, VIRTIO_NET_F_RSS);
if (n->has_vnet_hdr) {
n->curr_guest_offloads =
@@ -1260,7 +1270,9 @@ static void virtio_net_disable_rss(VirtIONet *n)
}
static uint16_t virtio_net_handle_rss(VirtIONet *n,
- struct iovec *iov, unsigned int iov_cnt)
+ struct iovec *iov,
+ unsigned int iov_cnt,
+ bool do_rss)
{
VirtIODevice *vdev = VIRTIO_DEVICE(n);
struct virtio_net_rss_config cfg;
@@ -1273,10 +1285,14 @@ static uint16_t virtio_net_handle_rss(VirtIONet *n,
const char *err_msg = "";
uint32_t err_value = 0;
- if (!virtio_vdev_has_feature(vdev, VIRTIO_NET_F_RSS)) {
+ if (do_rss && !virtio_vdev_has_feature(vdev, VIRTIO_NET_F_RSS)) {
err_msg = "RSS is not negotiated";
goto error;
}
+ if (!do_rss && !virtio_vdev_has_feature(vdev, VIRTIO_NET_F_HASH_REPORT)) {
+ err_msg = "Hash report is not negotiated";
+ goto error;
+ }
size_get = offsetof(struct virtio_net_rss_config, indirection_table);
s = iov_to_buf(iov, iov_cnt, offset, &cfg, size_get);
if (s != size_get) {
@@ -1288,6 +1304,9 @@ static uint16_t virtio_net_handle_rss(VirtIONet *n,
n->rss_data.indirections_len =
virtio_lduw_p(vdev, &cfg.indirection_table_mask);
n->rss_data.indirections_len++;
+ if (!do_rss) {
+ n->rss_data.indirections_len = 1;
+ }
if (!is_power_of_2(n->rss_data.indirections_len)) {
err_msg = "Invalid size of indirection table";
err_value = n->rss_data.indirections_len;
@@ -1298,8 +1317,8 @@ static uint16_t virtio_net_handle_rss(VirtIONet *n,
err_value = n->rss_data.indirections_len;
goto error;
}
- n->rss_data.default_queue =
- virtio_lduw_p(vdev, &cfg.unclassified_queue);
+ n->rss_data.default_queue = do_rss ?
+ virtio_lduw_p(vdev, &cfg.unclassified_queue) : 0;
if (n->rss_data.default_queue >= n->max_queues) {
err_msg = "Invalid default queue";
err_value = n->rss_data.default_queue;
@@ -1315,7 +1334,7 @@ static uint16_t virtio_net_handle_rss(VirtIONet *n,
}
for (i = 0; i < n->rss_data.indirections_len; ++i) {
uint16_t val = n->rss_data.indirections[i];
- n->rss_data.indirections[i] = virtio_lduw_p(vdev, &val);
+ n->rss_data.indirections[i] = do_rss ? virtio_lduw_p(vdev, &val) : 0;
}
offset += size_get;
size_get = sizeof(temp);
@@ -1325,7 +1344,7 @@ static uint16_t virtio_net_handle_rss(VirtIONet *n,
err_value = (uint32_t)s;
goto error;
}
- queues = virtio_lduw_p(vdev, &temp.us);
+ queues = do_rss ? virtio_lduw_p(vdev, &temp.us) : n->curr_queues;
if (queues == 0 || queues > n->max_queues) {
err_msg = "Invalid number of queues";
err_value = queues;
@@ -1371,8 +1390,12 @@ static int virtio_net_handle_mq(VirtIONet *n, uint8_t cmd,
uint16_t queues;
virtio_net_disable_rss(n);
+ if (cmd == VIRTIO_NET_CTRL_MQ_HASH_CONFIG) {
+ queues = virtio_net_handle_rss(n, iov, iov_cnt, false);
+ return queues ? VIRTIO_NET_OK : VIRTIO_NET_ERR;
+ }
if (cmd == VIRTIO_NET_CTRL_MQ_RSS_CONFIG) {
- queues = virtio_net_handle_rss(n, iov, iov_cnt);
+ queues = virtio_net_handle_rss(n, iov, iov_cnt, true);
} else if (cmd == VIRTIO_NET_CTRL_MQ_VQ_PAIRS_SET) {
struct virtio_net_ctrl_mq mq;
size_t s;
@@ -1659,6 +1682,14 @@ static uint8_t virtio_net_get_hash_type(bool isip4,
return 0xff;
}
+static void virtio_set_packet_hash(const uint8_t *buf, uint8_t report,
+ uint32_t hash)
+{
+ struct virtio_net_hdr_v1_hash *hdr = (void *)buf;
+ hdr->hash_value = hash;
+ hdr->hash_report = report;
+}
+
static int virtio_net_process_rss(NetClientState *nc, const uint8_t *buf,
size_t size)
{
@@ -1668,6 +1699,17 @@ static int virtio_net_process_rss(NetClientState *nc, const uint8_t *buf,
uint8_t net_hash_type;
uint32_t hash;
bool isip4, isip6, isudp, istcp;
+ static const uint8_t reports[NetPktRssIpV6UdpEx + 1] = {
+ VIRTIO_NET_HASH_REPORT_IPv4,
+ VIRTIO_NET_HASH_REPORT_TCPv4,
+ VIRTIO_NET_HASH_REPORT_TCPv6,
+ VIRTIO_NET_HASH_REPORT_IPv6,
+ VIRTIO_NET_HASH_REPORT_IPv6_EX,
+ VIRTIO_NET_HASH_REPORT_TCPv6_EX,
+ VIRTIO_NET_HASH_REPORT_UDPv4,
+ VIRTIO_NET_HASH_REPORT_UDPv6,
+ VIRTIO_NET_HASH_REPORT_UDPv6_EX
+ };
net_rx_pkt_set_protocols(pkt, buf + n->host_hdr_len,
size - n->host_hdr_len);
@@ -1681,12 +1723,25 @@ static int virtio_net_process_rss(NetClientState *nc, const uint8_t *buf,
net_hash_type = virtio_net_get_hash_type(isip4, isip6, isudp, istcp,
n->rss_data.hash_types);
if (net_hash_type > NetPktRssIpV6UdpEx) {
- return n->rss_data.default_queue;
+ if (n->rss_data.populate_hash) {
+ virtio_set_packet_hash(buf, VIRTIO_NET_HASH_REPORT_NONE, 0);
+ }
+ return n->rss_data.redirect ? n->rss_data.default_queue : -1;
}
hash = net_rx_pkt_calc_rss_hash(pkt, net_hash_type, n->rss_data.key);
- new_index = hash & (n->rss_data.indirections_len - 1);
- new_index = n->rss_data.indirections[new_index];
+
+ if (n->rss_data.populate_hash) {
+ virtio_set_packet_hash(buf, reports[net_hash_type], hash);
+ }
+
+ if (n->rss_data.redirect) {
+ new_index = hash & (n->rss_data.indirections_len - 1);
+ new_index = n->rss_data.indirections[new_index];
+ } else {
+ new_index = index;
+ }
+
if (index == new_index) {
return -1;
}
@@ -1766,6 +1821,11 @@ static ssize_t virtio_net_receive_rcu(NetClientState *nc, const uint8_t *buf,
}
receive_header(n, sg, elem->in_num, buf, size);
+ if (n->rss_data.populate_hash) {
+ offset = sizeof(mhdr);
+ iov_from_buf(sg, elem->in_num, offset,
+ buf + offset, n->host_hdr_len - sizeof(mhdr));
+ }
offset = n->host_hdr_len;
total += n->guest_hdr_len;
guest_offset = n->guest_hdr_len;
@@ -2758,7 +2818,9 @@ static int virtio_net_post_load_device(void *opaque, int version_id)
trace_virtio_net_post_load_device();
virtio_net_set_mrg_rx_bufs(n, n->mergeable_rx_bufs,
virtio_vdev_has_feature(vdev,
- VIRTIO_F_VERSION_1));
+ VIRTIO_F_VERSION_1),
+ virtio_vdev_has_feature(vdev,
+ VIRTIO_NET_F_HASH_REPORT));
/* MAC_TABLE_ENTRIES may be different from the saved image */
if (n->mac_table.in_use > MAC_TABLE_ENTRIES) {
@@ -3374,7 +3436,7 @@ static void virtio_net_device_realize(DeviceState *dev, Error **errp)
n->vqs[0].tx_waiting = 0;
n->tx_burst = n->net_conf.txburst;
- virtio_net_set_mrg_rx_bufs(n, 0, 0);
+ virtio_net_set_mrg_rx_bufs(n, 0, 0, 0);
n->promisc = 1; /* for compatibility */
n->mac_table.macs = g_malloc0(MAC_TABLE_ENTRIES * ETH_ALEN);
@@ -3528,6 +3590,8 @@ static Property virtio_net_properties[] = {
DEFINE_PROP_BIT64("mq", VirtIONet, host_features, VIRTIO_NET_F_MQ, false),
DEFINE_PROP_BIT64("rss", VirtIONet, host_features,
VIRTIO_NET_F_RSS, false),
+ DEFINE_PROP_BIT64("hash", VirtIONet, host_features,
+ VIRTIO_NET_F_HASH_REPORT, false),
DEFINE_PROP_BIT64("guest_rsc_ext", VirtIONet, host_features,
VIRTIO_NET_F_RSC_EXT, false),
DEFINE_PROP_UINT32("rsc_interval", VirtIONet, rsc_timeout,
@@ -131,6 +131,8 @@ typedef struct VirtioNetRscChain {
typedef struct VirtioNetRssData {
bool enabled;
+ bool redirect;
+ bool populate_hash;
uint32_t hash_types;
uint8_t key[VIRTIO_NET_RSS_MAX_KEY_SIZE];
uint16_t indirections[VIRTIO_NET_RSS_MAX_TABLE_LEN];
Suggest VIRTIO_NET_F_HASH_REPORT if specified in device parameters. If the VIRTIO_NET_F_HASH_REPORT is set, the device extends configuration space. If the feature is negotiated, the packet layout is extended to accomodate the hash information. In this case deliver packet's hash value and report type in virtio header extension. Use for configuration the same procedure as already used for RSS. We add two fields in rss_data that controls what the device does with the calculated hash if rss_data.enabled is set. If field 'populate' is set the hash is set in the packet, if field 'redirect' is set the hash is used to decide the queue to place the packet to. Signed-off-by: Yuri Benditovich <yuri.benditovich@daynix.com> --- hw/net/virtio-net.c | 98 ++++++++++++++++++++++++++++------ include/hw/virtio/virtio-net.h | 2 + 2 files changed, 83 insertions(+), 17 deletions(-)