diff options
Diffstat (limited to 'src/network/netdev')
28 files changed, 4046 insertions, 0 deletions
diff --git a/src/network/netdev/.gitignore b/src/network/netdev/.gitignore new file mode 100644 index 0000000000..0f1a65d2e6 --- /dev/null +++ b/src/network/netdev/.gitignore @@ -0,0 +1 @@ +/netdev-gperf.c diff --git a/src/network/netdev/bond.c b/src/network/netdev/bond.c new file mode 100644 index 0000000000..19b0e8da40 --- /dev/null +++ b/src/network/netdev/bond.c @@ -0,0 +1,444 @@ +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + Copyright 2014 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <netinet/ether.h> +#include <linux/if_bonding.h> + +#include "sd-netlink.h" + +#include "alloc-util.h" +#include "conf-parser.h" +#include "extract-word.h" +#include "missing.h" +#include "netdev/bond.h" +#include "string-table.h" +#include "string-util.h" + +/* + * Number of seconds between instances where the bonding + * driver sends learning packets to each slaves peer switch + */ +#define LEARNING_PACKETS_INTERVAL_MIN_SEC (1 * USEC_PER_SEC) +#define LEARNING_PACKETS_INTERVAL_MAX_SEC (0x7fffffff * USEC_PER_SEC) + +/* Number of IGMP membership reports to be issued after + * a failover event. + */ +#define RESEND_IGMP_MIN 0 +#define RESEND_IGMP_MAX 255 +#define RESEND_IGMP_DEFAULT 1 + +/* + * Number of packets to transmit through a slave before + * moving to the next one. + */ +#define PACKETS_PER_SLAVE_MIN 0 +#define PACKETS_PER_SLAVE_MAX 65535 +#define PACKETS_PER_SLAVE_DEFAULT 1 + +/* + * Number of peer notifications (gratuitous ARPs and + * unsolicited IPv6 Neighbor Advertisements) to be issued after a + * failover event. + */ +#define GRATUITOUS_ARP_MIN 0 +#define GRATUITOUS_ARP_MAX 255 +#define GRATUITOUS_ARP_DEFAULT 1 + +static const char* const bond_mode_table[_NETDEV_BOND_MODE_MAX] = { + [NETDEV_BOND_MODE_BALANCE_RR] = "balance-rr", + [NETDEV_BOND_MODE_ACTIVE_BACKUP] = "active-backup", + [NETDEV_BOND_MODE_BALANCE_XOR] = "balance-xor", + [NETDEV_BOND_MODE_BROADCAST] = "broadcast", + [NETDEV_BOND_MODE_802_3AD] = "802.3ad", + [NETDEV_BOND_MODE_BALANCE_TLB] = "balance-tlb", + [NETDEV_BOND_MODE_BALANCE_ALB] = "balance-alb", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_mode, BondMode); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_mode, bond_mode, BondMode, "Failed to parse bond mode"); + +static const char* const bond_xmit_hash_policy_table[_NETDEV_BOND_XMIT_HASH_POLICY_MAX] = { + [NETDEV_BOND_XMIT_HASH_POLICY_LAYER2] = "layer2", + [NETDEV_BOND_XMIT_HASH_POLICY_LAYER34] = "layer3+4", + [NETDEV_BOND_XMIT_HASH_POLICY_LAYER23] = "layer2+3", + [NETDEV_BOND_XMIT_HASH_POLICY_ENCAP23] = "encap2+3", + [NETDEV_BOND_XMIT_HASH_POLICY_ENCAP34] = "encap3+4", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_xmit_hash_policy, BondXmitHashPolicy); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_xmit_hash_policy, + bond_xmit_hash_policy, + BondXmitHashPolicy, + "Failed to parse bond transmit hash policy") + +static const char* const bond_lacp_rate_table[_NETDEV_BOND_LACP_RATE_MAX] = { + [NETDEV_BOND_LACP_RATE_SLOW] = "slow", + [NETDEV_BOND_LACP_RATE_FAST] = "fast", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_lacp_rate, BondLacpRate); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_lacp_rate, bond_lacp_rate, BondLacpRate, "Failed to parse bond lacp rate") + +static const char* const bond_ad_select_table[_NETDEV_BOND_AD_SELECT_MAX] = { + [NETDEV_BOND_AD_SELECT_STABLE] = "stable", + [NETDEV_BOND_AD_SELECT_BANDWIDTH] = "bandwidth", + [NETDEV_BOND_AD_SELECT_COUNT] = "count", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_ad_select, BondAdSelect); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_ad_select, bond_ad_select, BondAdSelect, "Failed to parse bond AD select"); + +static const char* const bond_fail_over_mac_table[_NETDEV_BOND_FAIL_OVER_MAC_MAX] = { + [NETDEV_BOND_FAIL_OVER_MAC_NONE] = "none", + [NETDEV_BOND_FAIL_OVER_MAC_ACTIVE] = "active", + [NETDEV_BOND_FAIL_OVER_MAC_FOLLOW] = "follow", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_fail_over_mac, BondFailOverMac); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_fail_over_mac, bond_fail_over_mac, BondFailOverMac, "Failed to parse bond fail over MAC"); + +static const char *const bond_arp_validate_table[_NETDEV_BOND_ARP_VALIDATE_MAX] = { + [NETDEV_BOND_ARP_VALIDATE_NONE] = "none", + [NETDEV_BOND_ARP_VALIDATE_ACTIVE]= "active", + [NETDEV_BOND_ARP_VALIDATE_BACKUP]= "backup", + [NETDEV_BOND_ARP_VALIDATE_ALL]= "all", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_arp_validate, BondArpValidate); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_arp_validate, bond_arp_validate, BondArpValidate, "Failed to parse bond arp validate"); + +static const char *const bond_arp_all_targets_table[_NETDEV_BOND_ARP_ALL_TARGETS_MAX] = { + [NETDEV_BOND_ARP_ALL_TARGETS_ANY] = "any", + [NETDEV_BOND_ARP_ALL_TARGETS_ALL] = "all", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_arp_all_targets, BondArpAllTargets); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_arp_all_targets, bond_arp_all_targets, BondArpAllTargets, "Failed to parse bond Arp all targets"); + +static const char *bond_primary_reselect_table[_NETDEV_BOND_PRIMARY_RESELECT_MAX] = { + [NETDEV_BOND_PRIMARY_RESELECT_ALWAYS] = "always", + [NETDEV_BOND_PRIMARY_RESELECT_BETTER]= "better", + [NETDEV_BOND_PRIMARY_RESELECT_FAILURE]= "failure", +}; + +DEFINE_STRING_TABLE_LOOKUP(bond_primary_reselect, BondPrimaryReselect); +DEFINE_CONFIG_PARSE_ENUM(config_parse_bond_primary_reselect, bond_primary_reselect, BondPrimaryReselect, "Failed to parse bond primary reselect"); + +static uint8_t bond_mode_to_kernel(BondMode mode) { + switch (mode) { + case NETDEV_BOND_MODE_BALANCE_RR: + return BOND_MODE_ROUNDROBIN; + case NETDEV_BOND_MODE_ACTIVE_BACKUP: + return BOND_MODE_ACTIVEBACKUP; + case NETDEV_BOND_MODE_BALANCE_XOR: + return BOND_MODE_XOR; + case NETDEV_BOND_MODE_BROADCAST: + return BOND_MODE_BROADCAST; + case NETDEV_BOND_MODE_802_3AD: + return BOND_MODE_8023AD; + case NETDEV_BOND_MODE_BALANCE_TLB: + return BOND_MODE_TLB; + case NETDEV_BOND_MODE_BALANCE_ALB: + return BOND_MODE_ALB; + default: + return (uint8_t) -1; + } +} + +static uint8_t bond_xmit_hash_policy_to_kernel(BondXmitHashPolicy policy) { + switch (policy) { + case NETDEV_BOND_XMIT_HASH_POLICY_LAYER2: + return BOND_XMIT_POLICY_LAYER2; + case NETDEV_BOND_XMIT_HASH_POLICY_LAYER34: + return BOND_XMIT_POLICY_LAYER34; + case NETDEV_BOND_XMIT_HASH_POLICY_LAYER23: + return BOND_XMIT_POLICY_LAYER23; + case NETDEV_BOND_XMIT_HASH_POLICY_ENCAP23: + return BOND_XMIT_POLICY_ENCAP23; + case NETDEV_BOND_XMIT_HASH_POLICY_ENCAP34: + return BOND_XMIT_POLICY_ENCAP34; + default: + return (uint8_t) -1; + } +} + +static int netdev_bond_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Bond *b; + ArpIpTarget *target = NULL; + int r, i = 0; + + assert(netdev); + assert(!link); + assert(m); + + b = BOND(netdev); + + assert(b); + + if (b->mode != _NETDEV_BOND_MODE_INVALID) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_MODE, + bond_mode_to_kernel(b->mode)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_MODE attribute: %m"); + } + + if (b->xmit_hash_policy != _NETDEV_BOND_XMIT_HASH_POLICY_INVALID) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_XMIT_HASH_POLICY, + bond_xmit_hash_policy_to_kernel(b->xmit_hash_policy)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_XMIT_HASH_POLICY attribute: %m"); + } + + if (b->lacp_rate != _NETDEV_BOND_LACP_RATE_INVALID && + b->mode == NETDEV_BOND_MODE_802_3AD) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_AD_LACP_RATE, b->lacp_rate ); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_AD_LACP_RATE attribute: %m"); + } + + if (b->miimon != 0) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_MIIMON, b->miimon / USEC_PER_MSEC); + if (r < 0) + log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_BOND_MIIMON attribute: %m"); + } + + if (b->downdelay != 0) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_DOWNDELAY, b->downdelay / USEC_PER_MSEC); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_DOWNDELAY attribute: %m"); + } + + if (b->updelay != 0) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_UPDELAY, b->updelay / USEC_PER_MSEC); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_UPDELAY attribute: %m"); + } + + if (b->arp_interval != 0) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_ARP_INTERVAL, b->arp_interval / USEC_PER_MSEC); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_ARP_INTERVAL attribute: %m"); + + if ((b->lp_interval >= LEARNING_PACKETS_INTERVAL_MIN_SEC) && + (b->lp_interval <= LEARNING_PACKETS_INTERVAL_MAX_SEC)) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_LP_INTERVAL, b->lp_interval / USEC_PER_SEC); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_LP_INTERVAL attribute: %m"); + } + } + + if (b->ad_select != _NETDEV_BOND_AD_SELECT_INVALID && + b->mode == NETDEV_BOND_MODE_802_3AD) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_AD_SELECT, b->ad_select); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_AD_SELECT attribute: %m"); + } + + if (b->fail_over_mac != _NETDEV_BOND_FAIL_OVER_MAC_INVALID && + b->mode == NETDEV_BOND_MODE_ACTIVE_BACKUP) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_FAIL_OVER_MAC, b->fail_over_mac); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_FAIL_OVER_MAC attribute: %m"); + } + + if (b->arp_validate != _NETDEV_BOND_ARP_VALIDATE_INVALID) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_ARP_VALIDATE, b->arp_validate); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_ARP_VALIDATE attribute: %m"); + } + + if (b->arp_all_targets != _NETDEV_BOND_ARP_ALL_TARGETS_INVALID) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_ARP_ALL_TARGETS, b->arp_all_targets); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_ARP_ALL_TARGETS attribute: %m"); + } + + if (b->primary_reselect != _NETDEV_BOND_PRIMARY_RESELECT_INVALID) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_PRIMARY_RESELECT, b->primary_reselect); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_PRIMARY_RESELECT attribute: %m"); + } + + if (b->resend_igmp <= RESEND_IGMP_MAX) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_RESEND_IGMP, b->resend_igmp); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_RESEND_IGMP attribute: %m"); + } + + if (b->packets_per_slave <= PACKETS_PER_SLAVE_MAX && + b->mode == NETDEV_BOND_MODE_BALANCE_RR) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_PACKETS_PER_SLAVE, b->packets_per_slave); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_PACKETS_PER_SLAVE attribute: %m"); + } + + if (b->num_grat_arp <= GRATUITOUS_ARP_MAX) { + r = sd_netlink_message_append_u8(m, IFLA_BOND_NUM_PEER_NOTIF, b->num_grat_arp); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_NUM_PEER_NOTIF attribute: %m"); + } + + if (b->min_links != 0) { + r = sd_netlink_message_append_u32(m, IFLA_BOND_MIN_LINKS, b->min_links); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_MIN_LINKS attribute: %m"); + } + + r = sd_netlink_message_append_u8(m, IFLA_BOND_ALL_SLAVES_ACTIVE, b->all_slaves_active); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_ALL_SLAVES_ACTIVE attribute: %m"); + + if (b->arp_interval > 0) { + if (b->n_arp_ip_targets > 0) { + + r = sd_netlink_message_open_container(m, IFLA_BOND_ARP_IP_TARGET); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not open contaniner IFLA_BOND_ARP_IP_TARGET : %m"); + + LIST_FOREACH(arp_ip_target, target, b->arp_ip_targets) { + r = sd_netlink_message_append_u32(m, i++, target->ip.in.s_addr); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BOND_ARP_ALL_TARGETS attribute: %m"); + } + + r = sd_netlink_message_close_container(m); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not close contaniner IFLA_BOND_ARP_IP_TARGET : %m"); + } + } + + return 0; +} + +int config_parse_arp_ip_target_address(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + Bond *b = userdata; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + for (;;) { + _cleanup_free_ ArpIpTarget *buffer = NULL; + _cleanup_free_ char *n = NULL; + int f; + + r = extract_first_word(&rvalue, &n, NULL, 0); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse Bond ARP ip target address, ignoring assignment: %s", rvalue); + return 0; + } + + if (r == 0) + break; + + buffer = new0(ArpIpTarget, 1); + if (!buffer) + return -ENOMEM; + + r = in_addr_from_string_auto(n, &f, &buffer->ip); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Bond ARP ip target address is invalid, ignoring assignment: %s", n); + return 0; + } + + if (f != AF_INET) { + log_syntax(unit, LOG_ERR, filename, line, 0, "Bond ARP ip target address is invalid, ignoring assignment: %s", n); + return 0; + } + + LIST_PREPEND(arp_ip_target, b->arp_ip_targets, buffer); + b->n_arp_ip_targets++; + + buffer = NULL; + } + + if (b->n_arp_ip_targets > NETDEV_BOND_ARP_TARGETS_MAX) + log_syntax(unit, LOG_WARNING, filename, line, 0, + "More than the maximum number of kernel-supported ARP ip targets specified: %d > %d", + b->n_arp_ip_targets, NETDEV_BOND_ARP_TARGETS_MAX); + + return 0; +} + +static void bond_done(NetDev *netdev) { + ArpIpTarget *t = NULL, *n = NULL; + Bond *b; + + assert(netdev); + + b = BOND(netdev); + + assert(b); + + LIST_FOREACH_SAFE(arp_ip_target, t, n, b->arp_ip_targets) + free(t); + + b->arp_ip_targets = NULL; +} + +static void bond_init(NetDev *netdev) { + Bond *b; + + assert(netdev); + + b = BOND(netdev); + + assert(b); + + b->mode = _NETDEV_BOND_MODE_INVALID; + b->xmit_hash_policy = _NETDEV_BOND_XMIT_HASH_POLICY_INVALID; + b->lacp_rate = _NETDEV_BOND_LACP_RATE_INVALID; + b->ad_select = _NETDEV_BOND_AD_SELECT_INVALID; + b->fail_over_mac = _NETDEV_BOND_FAIL_OVER_MAC_INVALID; + b->arp_validate = _NETDEV_BOND_ARP_VALIDATE_INVALID; + b->arp_all_targets = _NETDEV_BOND_ARP_ALL_TARGETS_INVALID; + b->primary_reselect = _NETDEV_BOND_PRIMARY_RESELECT_INVALID; + + b->all_slaves_active = false; + + b->resend_igmp = RESEND_IGMP_DEFAULT; + b->packets_per_slave = PACKETS_PER_SLAVE_DEFAULT; + b->num_grat_arp = GRATUITOUS_ARP_DEFAULT; + b->lp_interval = LEARNING_PACKETS_INTERVAL_MIN_SEC; + + LIST_HEAD_INIT(b->arp_ip_targets); + b->n_arp_ip_targets = 0; +} + +const NetDevVTable bond_vtable = { + .object_size = sizeof(Bond), + .init = bond_init, + .done = bond_done, + .sections = "Match\0NetDev\0Bond\0", + .fill_message_create = netdev_bond_fill_message_create, + .create_type = NETDEV_CREATE_MASTER, +}; diff --git a/src/network/netdev/bond.h b/src/network/netdev/bond.h new file mode 100644 index 0000000000..fb88b538ed --- /dev/null +++ b/src/network/netdev/bond.h @@ -0,0 +1,172 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "in-addr-util.h" +#include "list.h" + +#include "netdev/netdev.h" + +/* + * Maximum number of targets supported by the kernel for a single + * bond netdev. + */ +#define NETDEV_BOND_ARP_TARGETS_MAX 16 + +typedef enum BondMode { + NETDEV_BOND_MODE_BALANCE_RR, + NETDEV_BOND_MODE_ACTIVE_BACKUP, + NETDEV_BOND_MODE_BALANCE_XOR, + NETDEV_BOND_MODE_BROADCAST, + NETDEV_BOND_MODE_802_3AD, + NETDEV_BOND_MODE_BALANCE_TLB, + NETDEV_BOND_MODE_BALANCE_ALB, + _NETDEV_BOND_MODE_MAX, + _NETDEV_BOND_MODE_INVALID = -1 +} BondMode; + +typedef enum BondXmitHashPolicy { + NETDEV_BOND_XMIT_HASH_POLICY_LAYER2, + NETDEV_BOND_XMIT_HASH_POLICY_LAYER34, + NETDEV_BOND_XMIT_HASH_POLICY_LAYER23, + NETDEV_BOND_XMIT_HASH_POLICY_ENCAP23, + NETDEV_BOND_XMIT_HASH_POLICY_ENCAP34, + _NETDEV_BOND_XMIT_HASH_POLICY_MAX, + _NETDEV_BOND_XMIT_HASH_POLICY_INVALID = -1 +} BondXmitHashPolicy; + +typedef enum BondLacpRate { + NETDEV_BOND_LACP_RATE_SLOW, + NETDEV_BOND_LACP_RATE_FAST, + _NETDEV_BOND_LACP_RATE_MAX, + _NETDEV_BOND_LACP_RATE_INVALID = -1, +} BondLacpRate; + +typedef enum BondAdSelect { + NETDEV_BOND_AD_SELECT_STABLE, + NETDEV_BOND_AD_SELECT_BANDWIDTH, + NETDEV_BOND_AD_SELECT_COUNT, + _NETDEV_BOND_AD_SELECT_MAX, + _NETDEV_BOND_AD_SELECT_INVALID = -1, +} BondAdSelect; + +typedef enum BondFailOverMac { + NETDEV_BOND_FAIL_OVER_MAC_NONE, + NETDEV_BOND_FAIL_OVER_MAC_ACTIVE, + NETDEV_BOND_FAIL_OVER_MAC_FOLLOW, + _NETDEV_BOND_FAIL_OVER_MAC_MAX, + _NETDEV_BOND_FAIL_OVER_MAC_INVALID = -1, +} BondFailOverMac; + +typedef enum BondArpValidate { + NETDEV_BOND_ARP_VALIDATE_NONE, + NETDEV_BOND_ARP_VALIDATE_ACTIVE, + NETDEV_BOND_ARP_VALIDATE_BACKUP, + NETDEV_BOND_ARP_VALIDATE_ALL, + _NETDEV_BOND_ARP_VALIDATE_MAX, + _NETDEV_BOND_ARP_VALIDATE_INVALID = -1, +} BondArpValidate; + +typedef enum BondArpAllTargets { + NETDEV_BOND_ARP_ALL_TARGETS_ANY, + NETDEV_BOND_ARP_ALL_TARGETS_ALL, + _NETDEV_BOND_ARP_ALL_TARGETS_MAX, + _NETDEV_BOND_ARP_ALL_TARGETS_INVALID = -1, +} BondArpAllTargets; + +typedef enum BondPrimaryReselect { + NETDEV_BOND_PRIMARY_RESELECT_ALWAYS, + NETDEV_BOND_PRIMARY_RESELECT_BETTER, + NETDEV_BOND_PRIMARY_RESELECT_FAILURE, + _NETDEV_BOND_PRIMARY_RESELECT_MAX, + _NETDEV_BOND_PRIMARY_RESELECT_INVALID = -1, +} BondPrimaryReselect; + +typedef struct ArpIpTarget { + union in_addr_union ip; + + LIST_FIELDS(struct ArpIpTarget, arp_ip_target); +} ArpIpTarget; + +typedef struct Bond { + NetDev meta; + + BondMode mode; + BondXmitHashPolicy xmit_hash_policy; + BondLacpRate lacp_rate; + BondAdSelect ad_select; + BondFailOverMac fail_over_mac; + BondArpValidate arp_validate; + BondArpAllTargets arp_all_targets; + BondPrimaryReselect primary_reselect; + + bool all_slaves_active; + + unsigned resend_igmp; + unsigned packets_per_slave; + unsigned num_grat_arp; + unsigned min_links; + + usec_t miimon; + usec_t updelay; + usec_t downdelay; + usec_t arp_interval; + usec_t lp_interval; + + int n_arp_ip_targets; + ArpIpTarget *arp_ip_targets; +} Bond; + +DEFINE_NETDEV_CAST(BOND, Bond); +extern const NetDevVTable bond_vtable; + +const char *bond_mode_to_string(BondMode d) _const_; +BondMode bond_mode_from_string(const char *d) _pure_; + +const char *bond_xmit_hash_policy_to_string(BondXmitHashPolicy d) _const_; +BondXmitHashPolicy bond_xmit_hash_policy_from_string(const char *d) _pure_; + +const char *bond_lacp_rate_to_string(BondLacpRate d) _const_; +BondLacpRate bond_lacp_rate_from_string(const char *d) _pure_; + +const char *bond_fail_over_mac_to_string(BondFailOverMac d) _const_; +BondFailOverMac bond_fail_over_mac_from_string(const char *d) _pure_; + +const char *bond_ad_select_to_string(BondAdSelect d) _const_; +BondAdSelect bond_ad_select_from_string(const char *d) _pure_; + +const char *bond_arp_validate_to_string(BondArpValidate d) _const_; +BondArpValidate bond_arp_validate_from_string(const char *d) _pure_; + +const char *bond_arp_all_targets_to_string(BondArpAllTargets d) _const_; +BondArpAllTargets bond_arp_all_targets_from_string(const char *d) _pure_; + +const char *bond_primary_reselect_to_string(BondPrimaryReselect d) _const_; +BondPrimaryReselect bond_primary_reselect_from_string(const char *d) _pure_; + +int config_parse_bond_mode(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_xmit_hash_policy(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_lacp_rate(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_ad_select(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_fail_over_mac(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_arp_validate(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_arp_all_targets(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_bond_primary_reselect(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); +int config_parse_arp_ip_target_address(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); diff --git a/src/network/netdev/bridge.c b/src/network/netdev/bridge.c new file mode 100644 index 0000000000..760aefd97e --- /dev/null +++ b/src/network/netdev/bridge.c @@ -0,0 +1,171 @@ +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + Copyright 2014 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "missing.h" +#include "netlink-util.h" +#include "networkd.h" +#include "netdev/bridge.h" + +/* callback for brige netdev's parameter set */ +static int netdev_bridge_set_handler(sd_netlink *rtnl, sd_netlink_message *m, void *userdata) { + _cleanup_netdev_unref_ NetDev *netdev = userdata; + int r; + + assert(netdev); + assert(m); + + r = sd_netlink_message_get_errno(m); + if (r < 0) { + log_netdev_warning_errno(netdev, r, "Bridge parameters could not be set: %m"); + return 1; + } + + log_netdev_debug(netdev, "Bridge parameters set success"); + + return 1; +} + +static int netdev_bridge_post_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + _cleanup_(sd_netlink_message_unrefp) sd_netlink_message *req = NULL; + Bridge *b; + int r; + + assert(netdev); + + b = BRIDGE(netdev); + + assert(b); + + r = sd_rtnl_message_new_link(netdev->manager->rtnl, &req, RTM_NEWLINK, netdev->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not allocate RTM_SETLINK message: %m"); + + r = sd_netlink_message_set_flags(req, NLM_F_REQUEST | NLM_F_ACK); + if (r < 0) + return log_link_error_errno(link, r, "Could not set netlink flags: %m"); + + r = sd_netlink_message_open_container(req, IFLA_LINKINFO); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_PROTINFO attribute: %m"); + + r = sd_netlink_message_open_container_union(req, IFLA_INFO_DATA, netdev_kind_to_string(netdev->kind)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_INFO_DATA attribute: %m"); + + /* convert to jiffes */ + if (b->forward_delay > 0) { + r = sd_netlink_message_append_u32(req, IFLA_BR_FORWARD_DELAY, usec_to_jiffies(b->forward_delay)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_FORWARD_DELAY attribute: %m"); + } + + if (b->hello_time > 0) { + r = sd_netlink_message_append_u32(req, IFLA_BR_HELLO_TIME, usec_to_jiffies(b->hello_time)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_HELLO_TIME attribute: %m"); + } + + if (b->max_age > 0) { + r = sd_netlink_message_append_u32(req, IFLA_BR_MAX_AGE, usec_to_jiffies(b->max_age)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_MAX_AGE attribute: %m"); + } + + if (b->ageing_time > 0) { + r = sd_netlink_message_append_u32(req, IFLA_BR_AGEING_TIME, usec_to_jiffies(b->ageing_time)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_AGEING_TIME attribute: %m"); + } + + if (b->priority > 0) { + r = sd_netlink_message_append_u16(req, IFLA_BR_PRIORITY, b->priority); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_PRIORITY attribute: %m"); + } + + if (b->default_pvid > 0) { + r = sd_netlink_message_append_u16(req, IFLA_BR_VLAN_DEFAULT_PVID, b->default_pvid); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_VLAN_DEFAULT_PVID attribute: %m"); + } + + if (b->mcast_querier >= 0) { + r = sd_netlink_message_append_u8(req, IFLA_BR_MCAST_QUERIER, b->mcast_querier); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_MCAST_QUERIER attribute: %m"); + } + + if (b->mcast_snooping >= 0) { + r = sd_netlink_message_append_u8(req, IFLA_BR_MCAST_SNOOPING, b->mcast_snooping); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_MCAST_SNOOPING attribute: %m"); + } + + if (b->vlan_filtering >= 0) { + r = sd_netlink_message_append_u8(req, IFLA_BR_VLAN_FILTERING, b->vlan_filtering); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_VLAN_FILTERING attribute: %m"); + } + + if (b->stp >= 0) { + r = sd_netlink_message_append_u32(req, IFLA_BR_STP_STATE, b->stp); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_BR_STP_STATE attribute: %m"); + } + + r = sd_netlink_message_close_container(req); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_LINKINFO attribute: %m"); + + r = sd_netlink_message_close_container(req); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_INFO_DATA attribute: %m"); + + r = sd_netlink_call_async(netdev->manager->rtnl, req, netdev_bridge_set_handler, netdev, 0, NULL); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not send rtnetlink message: %m"); + + netdev_ref(netdev); + + return r; +} + +static void bridge_init(NetDev *n) { + Bridge *b; + + b = BRIDGE(n); + + assert(b); + + b->mcast_querier = -1; + b->mcast_snooping = -1; + b->vlan_filtering = -1; + b->stp = -1; +} + +const NetDevVTable bridge_vtable = { + .object_size = sizeof(Bridge), + .init = bridge_init, + .sections = "Match\0NetDev\0Bridge\0", + .post_create = netdev_bridge_post_create, + .create_type = NETDEV_CREATE_MASTER, +}; diff --git a/src/network/netdev/bridge.h b/src/network/netdev/bridge.h new file mode 100644 index 0000000000..093c60d5b5 --- /dev/null +++ b/src/network/netdev/bridge.h @@ -0,0 +1,41 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "netdev/netdev.h" + +typedef struct Bridge { + NetDev meta; + + int mcast_querier; + int mcast_snooping; + int vlan_filtering; + int stp; + uint16_t priority; + uint16_t default_pvid; + + usec_t forward_delay; + usec_t hello_time; + usec_t max_age; + usec_t ageing_time; +} Bridge; + +DEFINE_NETDEV_CAST(BRIDGE, Bridge); +extern const NetDevVTable bridge_vtable; diff --git a/src/network/netdev/dummy.c b/src/network/netdev/dummy.c new file mode 100644 index 0000000000..5e6e162931 --- /dev/null +++ b/src/network/netdev/dummy.c @@ -0,0 +1,28 @@ +/*** + This file is part of systemd. + + Copyright 2014 Susant Sahani <susant@redhat.com> + Copyright 2014 Tom Gundersen + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + + +#include "netdev/dummy.h" + +const NetDevVTable dummy_vtable = { + .object_size = sizeof(Dummy), + .sections = "Match\0NetDev\0", + .create_type = NETDEV_CREATE_INDEPENDENT, +}; diff --git a/src/network/netdev/dummy.h b/src/network/netdev/dummy.h new file mode 100644 index 0000000000..a908400459 --- /dev/null +++ b/src/network/netdev/dummy.h @@ -0,0 +1,29 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "netdev/netdev.h" + +typedef struct Dummy { + NetDev meta; +} Dummy; + +DEFINE_NETDEV_CAST(DUMMY, Dummy); +extern const NetDevVTable dummy_vtable; diff --git a/src/network/netdev/ipvlan.c b/src/network/netdev/ipvlan.c new file mode 100644 index 0000000000..3b5c30fed8 --- /dev/null +++ b/src/network/netdev/ipvlan.c @@ -0,0 +1,73 @@ +/*** + This file is part of systemd. + + Copyright 2013-2015 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "conf-parser.h" +#include "netdev/ipvlan.h" +#include "string-table.h" + +static const char* const ipvlan_mode_table[_NETDEV_IPVLAN_MODE_MAX] = { + [NETDEV_IPVLAN_MODE_L2] = "L2", + [NETDEV_IPVLAN_MODE_L3] = "L3", +}; + +DEFINE_STRING_TABLE_LOOKUP(ipvlan_mode, IPVlanMode); +DEFINE_CONFIG_PARSE_ENUM(config_parse_ipvlan_mode, ipvlan_mode, IPVlanMode, "Failed to parse ipvlan mode"); + +static int netdev_ipvlan_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *req) { + IPVlan *m; + int r; + + assert(netdev); + assert(link); + assert(netdev->ifname); + + m = IPVLAN(netdev); + + assert(m); + + if (m->mode != _NETDEV_IPVLAN_MODE_INVALID) { + r = sd_netlink_message_append_u16(req, IFLA_IPVLAN_MODE, m->mode); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPVLAN_MODE attribute: %m"); + } + + return 0; +} + +static void ipvlan_init(NetDev *n) { + IPVlan *m; + + assert(n); + + m = IPVLAN(n); + + assert(m); + + m->mode = _NETDEV_IPVLAN_MODE_INVALID; +} + +const NetDevVTable ipvlan_vtable = { + .object_size = sizeof(IPVlan), + .init = ipvlan_init, + .sections = "Match\0NetDev\0IPVLAN\0", + .fill_message_create = netdev_ipvlan_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, +}; diff --git a/src/network/netdev/ipvlan.h b/src/network/netdev/ipvlan.h new file mode 100644 index 0000000000..7d7d0184f1 --- /dev/null +++ b/src/network/netdev/ipvlan.h @@ -0,0 +1,44 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014-2015 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "missing.h" +#include "netdev/netdev.h" + +typedef enum IPVlanMode { + NETDEV_IPVLAN_MODE_L2 = IPVLAN_MODE_L2, + NETDEV_IPVLAN_MODE_L3 = IPVLAN_MODE_L3, + _NETDEV_IPVLAN_MODE_MAX, + _NETDEV_IPVLAN_MODE_INVALID = -1 +} IPVlanMode; + +typedef struct IPVlan { + NetDev meta; + + IPVlanMode mode; +} IPVlan; + +DEFINE_NETDEV_CAST(IPVLAN, IPVlan); +extern const NetDevVTable ipvlan_vtable; + +const char *ipvlan_mode_to_string(IPVlanMode d) _const_; +IPVlanMode ipvlan_mode_from_string(const char *d) _pure_; + +int config_parse_ipvlan_mode(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); diff --git a/src/network/netdev/macvlan.c b/src/network/netdev/macvlan.c new file mode 100644 index 0000000000..93f650def5 --- /dev/null +++ b/src/network/netdev/macvlan.c @@ -0,0 +1,89 @@ +/*** + This file is part of systemd. + + Copyright 2013 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "conf-parser.h" +#include "netdev/macvlan.h" +#include "string-table.h" + +static const char* const macvlan_mode_table[_NETDEV_MACVLAN_MODE_MAX] = { + [NETDEV_MACVLAN_MODE_PRIVATE] = "private", + [NETDEV_MACVLAN_MODE_VEPA] = "vepa", + [NETDEV_MACVLAN_MODE_BRIDGE] = "bridge", + [NETDEV_MACVLAN_MODE_PASSTHRU] = "passthru", +}; + +DEFINE_STRING_TABLE_LOOKUP(macvlan_mode, MacVlanMode); +DEFINE_CONFIG_PARSE_ENUM(config_parse_macvlan_mode, macvlan_mode, MacVlanMode, "Failed to parse macvlan mode"); + +static int netdev_macvlan_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *req) { + MacVlan *m; + int r; + + assert(netdev); + assert(link); + assert(netdev->ifname); + + if (netdev->kind == NETDEV_KIND_MACVLAN) + m = MACVLAN(netdev); + else + m = MACVTAP(netdev); + + assert(m); + + if (m->mode != _NETDEV_MACVLAN_MODE_INVALID) { + r = sd_netlink_message_append_u32(req, IFLA_MACVLAN_MODE, m->mode); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_MACVLAN_MODE attribute: %m"); + } + + return 0; +} + +static void macvlan_init(NetDev *n) { + MacVlan *m; + + assert(n); + + if (n->kind == NETDEV_KIND_MACVLAN) + m = MACVLAN(n); + else + m = MACVTAP(n); + + assert(m); + + m->mode = _NETDEV_MACVLAN_MODE_INVALID; +} + +const NetDevVTable macvtap_vtable = { + .object_size = sizeof(MacVlan), + .init = macvlan_init, + .sections = "Match\0NetDev\0MACVTAP\0", + .fill_message_create = netdev_macvlan_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, +}; + +const NetDevVTable macvlan_vtable = { + .object_size = sizeof(MacVlan), + .init = macvlan_init, + .sections = "Match\0NetDev\0MACVLAN\0", + .fill_message_create = netdev_macvlan_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, +}; diff --git a/src/network/netdev/macvlan.h b/src/network/netdev/macvlan.h new file mode 100644 index 0000000000..118d55658c --- /dev/null +++ b/src/network/netdev/macvlan.h @@ -0,0 +1,49 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct MacVlan MacVlan; + +#include "netdev/netdev.h" + +typedef enum MacVlanMode { + NETDEV_MACVLAN_MODE_PRIVATE = MACVLAN_MODE_PRIVATE, + NETDEV_MACVLAN_MODE_VEPA = MACVLAN_MODE_VEPA, + NETDEV_MACVLAN_MODE_BRIDGE = MACVLAN_MODE_BRIDGE, + NETDEV_MACVLAN_MODE_PASSTHRU = MACVLAN_MODE_PASSTHRU, + _NETDEV_MACVLAN_MODE_MAX, + _NETDEV_MACVLAN_MODE_INVALID = -1 +} MacVlanMode; + +struct MacVlan { + NetDev meta; + + MacVlanMode mode; +}; + +DEFINE_NETDEV_CAST(MACVLAN, MacVlan); +DEFINE_NETDEV_CAST(MACVTAP, MacVlan); +extern const NetDevVTable macvlan_vtable; +extern const NetDevVTable macvtap_vtable; + +const char *macvlan_mode_to_string(MacVlanMode d) _const_; +MacVlanMode macvlan_mode_from_string(const char *d) _pure_; + +int config_parse_macvlan_mode(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); diff --git a/src/network/netdev/netdev-gperf.gperf b/src/network/netdev/netdev-gperf.gperf new file mode 100644 index 0000000000..b3461e39a9 --- /dev/null +++ b/src/network/netdev/netdev-gperf.gperf @@ -0,0 +1,118 @@ +%{ +#include <stddef.h> +#include "conf-parser.h" +#include "network-internal.h" +#include "netdev/bond.h" +#include "netdev/bridge.h" +#include "netdev/ipvlan.h" +#include "netdev/macvlan.h" +#include "netdev/tunnel.h" +#include "netdev/tuntap.h" +#include "netdev/veth.h" +#include "netdev/vlan.h" +#include "netdev/vxlan.h" +#include "netdev/vrf.h" +#include "netdev/netdev.h" +#include "vlan-util.h" +%} +struct ConfigPerfItem; +%null_strings +%language=ANSI-C +%define slot-name section_and_lvalue +%define hash-function-name network_netdev_gperf_hash +%define lookup-function-name network_netdev_gperf_lookup +%readonly-tables +%omit-struct-type +%struct-type +%includes +%% +Match.Host, config_parse_net_condition, CONDITION_HOST, offsetof(NetDev, match_host) +Match.Virtualization, config_parse_net_condition, CONDITION_VIRTUALIZATION, offsetof(NetDev, match_virt) +Match.KernelCommandLine, config_parse_net_condition, CONDITION_KERNEL_COMMAND_LINE, offsetof(NetDev, match_kernel) +Match.Architecture, config_parse_net_condition, CONDITION_ARCHITECTURE, offsetof(NetDev, match_arch) +NetDev.Description, config_parse_string, 0, offsetof(NetDev, description) +NetDev.Name, config_parse_ifname, 0, offsetof(NetDev, ifname) +NetDev.Kind, config_parse_netdev_kind, 0, offsetof(NetDev, kind) +NetDev.MTUBytes, config_parse_iec_size, 0, offsetof(NetDev, mtu) +NetDev.MACAddress, config_parse_hwaddr, 0, offsetof(NetDev, mac) +VLAN.Id, config_parse_vlanid, 0, offsetof(VLan, id) +MACVLAN.Mode, config_parse_macvlan_mode, 0, offsetof(MacVlan, mode) +MACVTAP.Mode, config_parse_macvlan_mode, 0, offsetof(MacVlan, mode) +IPVLAN.Mode, config_parse_ipvlan_mode, 0, offsetof(IPVlan, mode) +Tunnel.Local, config_parse_tunnel_address, 0, offsetof(Tunnel, local) +Tunnel.Remote, config_parse_tunnel_address, 0, offsetof(Tunnel, remote) +Tunnel.TOS, config_parse_unsigned, 0, offsetof(Tunnel, tos) +Tunnel.TTL, config_parse_unsigned, 0, offsetof(Tunnel, ttl) +Tunnel.Key, config_parse_tunnel_key, 0, offsetof(Tunnel, key) +Tunnel.InputKey, config_parse_tunnel_key, 0, offsetof(Tunnel, ikey) +Tunnel.OutputKey, config_parse_tunnel_key, 0, offsetof(Tunnel, okey) +Tunnel.DiscoverPathMTU, config_parse_bool, 0, offsetof(Tunnel, pmtudisc) +Tunnel.Mode, config_parse_ip6tnl_mode, 0, offsetof(Tunnel, ip6tnl_mode) +Tunnel.IPv6FlowLabel, config_parse_ipv6_flowlabel, 0, offsetof(Tunnel, ipv6_flowlabel) +Tunnel.CopyDSCP, config_parse_bool, 0, offsetof(Tunnel, copy_dscp) +Tunnel.EncapsulationLimit, config_parse_encap_limit, 0, offsetof(Tunnel, encap_limit) +Peer.Name, config_parse_ifname, 0, offsetof(Veth, ifname_peer) +Peer.MACAddress, config_parse_hwaddr, 0, offsetof(Veth, mac_peer) +VXLAN.Id, config_parse_uint64, 0, offsetof(VxLan, id) +VXLAN.Group, config_parse_vxlan_group_address, 0, offsetof(VxLan, group) +VXLAN.TOS, config_parse_unsigned, 0, offsetof(VxLan, tos) +VXLAN.TTL, config_parse_unsigned, 0, offsetof(VxLan, ttl) +VXLAN.MacLearning, config_parse_bool, 0, offsetof(VxLan, learning) +VXLAN.ARPProxy, config_parse_bool, 0, offsetof(VxLan, arp_proxy) +VXLAN.L2MissNotification, config_parse_bool, 0, offsetof(VxLan, l2miss) +VXLAN.L3MissNotification, config_parse_bool, 0, offsetof(VxLan, l3miss) +VXLAN.RouteShortCircuit, config_parse_bool, 0, offsetof(VxLan, route_short_circuit) +VXLAN.UDPCheckSum, config_parse_bool, 0, offsetof(VxLan, udpcsum) +VXLAN.UDPChecksum, config_parse_bool, 0, offsetof(VxLan, udpcsum) +VXLAN.UDP6ZeroCheckSumRx, config_parse_bool, 0, offsetof(VxLan, udp6zerocsumrx) +VXLAN.UDP6ZeroChecksumRx, config_parse_bool, 0, offsetof(VxLan, udp6zerocsumrx) +VXLAN.UDP6ZeroCheckSumTx, config_parse_bool, 0, offsetof(VxLan, udp6zerocsumtx) +VXLAN.UDP6ZeroChecksumTx, config_parse_bool, 0, offsetof(VxLan, udp6zerocsumtx) +VXLAN.RemoteChecksumTx, config_parse_bool, 0, offsetof(VxLan, remote_csum_tx) +VXLAN.RemoteChecksumRx, config_parse_bool, 0, offsetof(VxLan, remote_csum_rx) +VXLAN.FDBAgeingSec, config_parse_sec, 0, offsetof(VxLan, fdb_ageing) +VXLAN.GroupPolicyExtension, config_parse_bool, 0, offsetof(VxLan, group_policy) +VXLAN.MaximumFDBEntries, config_parse_unsigned, 0, offsetof(VxLan, max_fdb) +VXLAN.PortRange, config_parse_port_range, 0, 0 +VXLAN.DestinationPort, config_parse_destination_port, 0, offsetof(VxLan, dest_port) +Tun.OneQueue, config_parse_bool, 0, offsetof(TunTap, one_queue) +Tun.MultiQueue, config_parse_bool, 0, offsetof(TunTap, multi_queue) +Tun.PacketInfo, config_parse_bool, 0, offsetof(TunTap, packet_info) +Tun.User, config_parse_string, 0, offsetof(TunTap, user_name) +Tun.Group, config_parse_string, 0, offsetof(TunTap, group_name) +Tap.OneQueue, config_parse_bool, 0, offsetof(TunTap, one_queue) +Tap.MultiQueue, config_parse_bool, 0, offsetof(TunTap, multi_queue) +Tap.PacketInfo, config_parse_bool, 0, offsetof(TunTap, packet_info) +Tap.VNetHeader, config_parse_bool, 0, offsetof(TunTap, vnet_hdr) +Tap.User, config_parse_string, 0, offsetof(TunTap, user_name) +Tap.Group, config_parse_string, 0, offsetof(TunTap, group_name) +Bond.Mode, config_parse_bond_mode, 0, offsetof(Bond, mode) +Bond.TransmitHashPolicy, config_parse_bond_xmit_hash_policy, 0, offsetof(Bond, xmit_hash_policy) +Bond.LACPTransmitRate, config_parse_bond_lacp_rate, 0, offsetof(Bond, lacp_rate) +Bond.AdSelect, config_parse_bond_ad_select, 0, offsetof(Bond, ad_select) +Bond.FailOverMACPolicy, config_parse_bond_fail_over_mac, 0, offsetof(Bond, fail_over_mac) +Bond.ARPIPTargets, config_parse_arp_ip_target_address, 0, 0 +Bond.ARPValidate, config_parse_bond_arp_validate, 0, offsetof(Bond, arp_validate) +Bond.ARPAllTargets, config_parse_bond_arp_all_targets, 0, offsetof(Bond, arp_all_targets) +Bond.PrimaryReselectPolicy, config_parse_bond_primary_reselect, 0, offsetof(Bond, primary_reselect) +Bond.ResendIGMP, config_parse_unsigned, 0, offsetof(Bond, resend_igmp) +Bond.PacketsPerSlave, config_parse_unsigned, 0, offsetof(Bond, packets_per_slave) +Bond.GratuitousARP, config_parse_unsigned, 0, offsetof(Bond, num_grat_arp) +Bond.AllSlavesActive, config_parse_unsigned, 0, offsetof(Bond, all_slaves_active) +Bond.MinLinks, config_parse_unsigned, 0, offsetof(Bond, min_links) +Bond.MIIMonitorSec, config_parse_sec, 0, offsetof(Bond, miimon) +Bond.UpDelaySec, config_parse_sec, 0, offsetof(Bond, updelay) +Bond.DownDelaySec, config_parse_sec, 0, offsetof(Bond, downdelay) +Bond.ARPIntervalSec, config_parse_sec, 0, offsetof(Bond, arp_interval) +Bond.LearnPacketIntervalSec, config_parse_sec, 0, offsetof(Bond, lp_interval) +Bridge.HelloTimeSec, config_parse_sec, 0, offsetof(Bridge, hello_time) +Bridge.MaxAgeSec, config_parse_sec, 0, offsetof(Bridge, max_age) +Bridge.AgeingTimeSec, config_parse_sec, 0, offsetof(Bridge, ageing_time) +Bridge.ForwardDelaySec, config_parse_sec, 0, offsetof(Bridge, forward_delay) +Bridge.Priority, config_parse_uint16, 0, offsetof(Bridge, priority) +Bridge.DefaultPVID, config_parse_vlanid, 0, offsetof(Bridge, default_pvid) +Bridge.MulticastQuerier, config_parse_tristate, 0, offsetof(Bridge, mcast_querier) +Bridge.MulticastSnooping, config_parse_tristate, 0, offsetof(Bridge, mcast_snooping) +Bridge.VLANFiltering, config_parse_tristate, 0, offsetof(Bridge, vlan_filtering) +Bridge.STP, config_parse_tristate, 0, offsetof(Bridge, stp) +VRF.TableId, config_parse_uint32, 0, offsetof(Vrf, table_id) diff --git a/src/network/netdev/netdev.c b/src/network/netdev/netdev.c new file mode 100644 index 0000000000..4f287bab72 --- /dev/null +++ b/src/network/netdev/netdev.c @@ -0,0 +1,717 @@ +/*** + This file is part of systemd. + + Copyright 2013 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "alloc-util.h" +#include "conf-files.h" +#include "conf-parser.h" +#include "fd-util.h" +#include "list.h" +#include "netlink-util.h" +#include "network-internal.h" +#include "netdev/netdev.h" +#include "networkd.h" +#include "siphash24.h" +#include "stat-util.h" +#include "string-table.h" +#include "string-util.h" + +const NetDevVTable * const netdev_vtable[_NETDEV_KIND_MAX] = { + [NETDEV_KIND_BRIDGE] = &bridge_vtable, + [NETDEV_KIND_BOND] = &bond_vtable, + [NETDEV_KIND_VLAN] = &vlan_vtable, + [NETDEV_KIND_MACVLAN] = &macvlan_vtable, + [NETDEV_KIND_MACVTAP] = &macvtap_vtable, + [NETDEV_KIND_IPVLAN] = &ipvlan_vtable, + [NETDEV_KIND_VXLAN] = &vxlan_vtable, + [NETDEV_KIND_IPIP] = &ipip_vtable, + [NETDEV_KIND_GRE] = &gre_vtable, + [NETDEV_KIND_GRETAP] = &gretap_vtable, + [NETDEV_KIND_IP6GRE] = &ip6gre_vtable, + [NETDEV_KIND_IP6GRETAP] = &ip6gretap_vtable, + [NETDEV_KIND_SIT] = &sit_vtable, + [NETDEV_KIND_VTI] = &vti_vtable, + [NETDEV_KIND_VTI6] = &vti6_vtable, + [NETDEV_KIND_VETH] = &veth_vtable, + [NETDEV_KIND_DUMMY] = &dummy_vtable, + [NETDEV_KIND_TUN] = &tun_vtable, + [NETDEV_KIND_TAP] = &tap_vtable, + [NETDEV_KIND_IP6TNL] = &ip6tnl_vtable, + [NETDEV_KIND_VRF] = &vrf_vtable, + [NETDEV_KIND_VCAN] = &vcan_vtable, +}; + +static const char* const netdev_kind_table[_NETDEV_KIND_MAX] = { + [NETDEV_KIND_BRIDGE] = "bridge", + [NETDEV_KIND_BOND] = "bond", + [NETDEV_KIND_VLAN] = "vlan", + [NETDEV_KIND_MACVLAN] = "macvlan", + [NETDEV_KIND_MACVTAP] = "macvtap", + [NETDEV_KIND_IPVLAN] = "ipvlan", + [NETDEV_KIND_VXLAN] = "vxlan", + [NETDEV_KIND_IPIP] = "ipip", + [NETDEV_KIND_GRE] = "gre", + [NETDEV_KIND_GRETAP] = "gretap", + [NETDEV_KIND_IP6GRE] = "ip6gre", + [NETDEV_KIND_IP6GRETAP] = "ip6gretap", + [NETDEV_KIND_SIT] = "sit", + [NETDEV_KIND_VETH] = "veth", + [NETDEV_KIND_VTI] = "vti", + [NETDEV_KIND_VTI6] = "vti6", + [NETDEV_KIND_DUMMY] = "dummy", + [NETDEV_KIND_TUN] = "tun", + [NETDEV_KIND_TAP] = "tap", + [NETDEV_KIND_IP6TNL] = "ip6tnl", + [NETDEV_KIND_VRF] = "vrf", + [NETDEV_KIND_VCAN] = "vcan", +}; + +DEFINE_STRING_TABLE_LOOKUP(netdev_kind, NetDevKind); +DEFINE_CONFIG_PARSE_ENUM(config_parse_netdev_kind, netdev_kind, NetDevKind, "Failed to parse netdev kind"); + +static void netdev_cancel_callbacks(NetDev *netdev) { + _cleanup_(sd_netlink_message_unrefp) sd_netlink_message *m = NULL; + netdev_join_callback *callback; + + if (!netdev) + return; + + rtnl_message_new_synthetic_error(-ENODEV, 0, &m); + + while ((callback = netdev->callbacks)) { + if (m) { + assert(callback->link); + assert(callback->callback); + assert(netdev->manager); + assert(netdev->manager->rtnl); + + callback->callback(netdev->manager->rtnl, m, callback->link); + } + + LIST_REMOVE(callbacks, netdev->callbacks, callback); + link_unref(callback->link); + free(callback); + } +} + +static void netdev_free(NetDev *netdev) { + if (!netdev) + return; + + netdev_cancel_callbacks(netdev); + + if (netdev->ifname) + hashmap_remove(netdev->manager->netdevs, netdev->ifname); + + free(netdev->filename); + + free(netdev->description); + free(netdev->ifname); + free(netdev->mac); + + condition_free_list(netdev->match_host); + condition_free_list(netdev->match_virt); + condition_free_list(netdev->match_kernel); + condition_free_list(netdev->match_arch); + + if (NETDEV_VTABLE(netdev) && + NETDEV_VTABLE(netdev)->done) + NETDEV_VTABLE(netdev)->done(netdev); + + free(netdev); +} + +NetDev *netdev_unref(NetDev *netdev) { + if (netdev && (-- netdev->n_ref <= 0)) + netdev_free(netdev); + + return NULL; +} + +NetDev *netdev_ref(NetDev *netdev) { + if (netdev) + assert_se(++ netdev->n_ref >= 2); + + return netdev; +} + +void netdev_drop(NetDev *netdev) { + if (!netdev || netdev->state == NETDEV_STATE_LINGER) + return; + + netdev->state = NETDEV_STATE_LINGER; + + log_netdev_debug(netdev, "netdev removed"); + + netdev_cancel_callbacks(netdev); + + netdev_unref(netdev); + + return; +} + +int netdev_get(Manager *manager, const char *name, NetDev **ret) { + NetDev *netdev; + + assert(manager); + assert(name); + assert(ret); + + netdev = hashmap_get(manager->netdevs, name); + if (!netdev) { + *ret = NULL; + return -ENOENT; + } + + *ret = netdev; + + return 0; +} + +static int netdev_enter_failed(NetDev *netdev) { + netdev->state = NETDEV_STATE_FAILED; + + netdev_cancel_callbacks(netdev); + + return 0; +} + +static int netdev_enslave_ready(NetDev *netdev, Link* link, sd_netlink_message_handler_t callback) { + _cleanup_(sd_netlink_message_unrefp) sd_netlink_message *req = NULL; + int r; + + assert(netdev); + assert(netdev->state == NETDEV_STATE_READY); + assert(netdev->manager); + assert(netdev->manager->rtnl); + assert(IN_SET(netdev->kind, NETDEV_KIND_BRIDGE, NETDEV_KIND_BOND, NETDEV_KIND_VRF)); + assert(link); + assert(callback); + + r = sd_rtnl_message_new_link(netdev->manager->rtnl, &req, RTM_SETLINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not allocate RTM_SETLINK message: %m"); + + r = sd_netlink_message_append_u32(req, IFLA_MASTER, netdev->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_MASTER attribute: %m"); + + r = sd_netlink_call_async(netdev->manager->rtnl, req, callback, link, 0, NULL); + if (r < 0) + return log_netdev_error(netdev, "Could not send rtnetlink message: %m"); + + link_ref(link); + + log_netdev_debug(netdev, "Enslaving link '%s'", link->ifname); + + return 0; +} + +static int netdev_enter_ready(NetDev *netdev) { + netdev_join_callback *callback, *callback_next; + int r; + + assert(netdev); + assert(netdev->ifname); + + if (netdev->state != NETDEV_STATE_CREATING) + return 0; + + netdev->state = NETDEV_STATE_READY; + + log_netdev_info(netdev, "netdev ready"); + + LIST_FOREACH_SAFE(callbacks, callback, callback_next, netdev->callbacks) { + /* enslave the links that were attempted to be enslaved before the + * link was ready */ + r = netdev_enslave_ready(netdev, callback->link, callback->callback); + if (r < 0) + return r; + + LIST_REMOVE(callbacks, netdev->callbacks, callback); + link_unref(callback->link); + free(callback); + } + + if (NETDEV_VTABLE(netdev)->post_create) + NETDEV_VTABLE(netdev)->post_create(netdev, NULL, NULL); + + return 0; +} + +/* callback for netdev's created without a backing Link */ +static int netdev_create_handler(sd_netlink *rtnl, sd_netlink_message *m, void *userdata) { + _cleanup_netdev_unref_ NetDev *netdev = userdata; + int r; + + assert(netdev->state != _NETDEV_STATE_INVALID); + + r = sd_netlink_message_get_errno(m); + if (r == -EEXIST) + log_netdev_info(netdev, "netdev exists, using existing without changing its parameters"); + else if (r < 0) { + log_netdev_warning_errno(netdev, r, "netdev could not be created: %m"); + netdev_drop(netdev); + + return 1; + } + + log_netdev_debug(netdev, "Created"); + + return 1; +} + +int netdev_enslave(NetDev *netdev, Link *link, sd_netlink_message_handler_t callback) { + int r; + + assert(netdev); + assert(netdev->manager); + assert(netdev->manager->rtnl); + assert(IN_SET(netdev->kind, NETDEV_KIND_BRIDGE, NETDEV_KIND_BOND, NETDEV_KIND_VRF)); + + if (netdev->state == NETDEV_STATE_READY) { + r = netdev_enslave_ready(netdev, link, callback); + if (r < 0) + return r; + } else if (IN_SET(netdev->state, NETDEV_STATE_LINGER, NETDEV_STATE_FAILED)) { + _cleanup_(sd_netlink_message_unrefp) sd_netlink_message *m = NULL; + + r = rtnl_message_new_synthetic_error(-ENODEV, 0, &m); + if (r >= 0) + callback(netdev->manager->rtnl, m, link); + } else { + /* the netdev is not yet read, save this request for when it is */ + netdev_join_callback *cb; + + cb = new0(netdev_join_callback, 1); + if (!cb) + return log_oom(); + + cb->callback = callback; + cb->link = link; + link_ref(link); + + LIST_PREPEND(callbacks, netdev->callbacks, cb); + + log_netdev_debug(netdev, "Will enslave '%s', when ready", link->ifname); + } + + return 0; +} + +int netdev_set_ifindex(NetDev *netdev, sd_netlink_message *message) { + uint16_t type; + const char *kind; + const char *received_kind; + const char *received_name; + int r, ifindex; + + assert(netdev); + assert(message); + + r = sd_netlink_message_get_type(message, &type); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not get rtnl message type: %m"); + + if (type != RTM_NEWLINK) { + log_netdev_error(netdev, "Cannot set ifindex from unexpected rtnl message type."); + return -EINVAL; + } + + r = sd_rtnl_message_link_get_ifindex(message, &ifindex); + if (r < 0) { + log_netdev_error_errno(netdev, r, "Could not get ifindex: %m"); + netdev_enter_failed(netdev); + return r; + } else if (ifindex <= 0) { + log_netdev_error(netdev, "Got invalid ifindex: %d", ifindex); + netdev_enter_failed(netdev); + return -EINVAL; + } + + if (netdev->ifindex > 0) { + if (netdev->ifindex != ifindex) { + log_netdev_error(netdev, "Could not set ifindex to %d, already set to %d", + ifindex, netdev->ifindex); + netdev_enter_failed(netdev); + return -EEXIST; + } else + /* ifindex already set to the same for this netdev */ + return 0; + } + + r = sd_netlink_message_read_string(message, IFLA_IFNAME, &received_name); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not get IFNAME: %m"); + + if (!streq(netdev->ifname, received_name)) { + log_netdev_error(netdev, "Received newlink with wrong IFNAME %s", received_name); + netdev_enter_failed(netdev); + return r; + } + + r = sd_netlink_message_enter_container(message, IFLA_LINKINFO); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not get LINKINFO: %m"); + + r = sd_netlink_message_read_string(message, IFLA_INFO_KIND, &received_kind); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not get KIND: %m"); + + r = sd_netlink_message_exit_container(message); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not exit container: %m"); + + if (netdev->kind == NETDEV_KIND_TAP) + /* the kernel does not distinguish between tun and tap */ + kind = "tun"; + else { + kind = netdev_kind_to_string(netdev->kind); + if (!kind) { + log_netdev_error(netdev, "Could not get kind"); + netdev_enter_failed(netdev); + return -EINVAL; + } + } + + if (!streq(kind, received_kind)) { + log_netdev_error(netdev, + "Received newlink with wrong KIND %s, " + "expected %s", received_kind, kind); + netdev_enter_failed(netdev); + return r; + } + + netdev->ifindex = ifindex; + + log_netdev_debug(netdev, "netdev has index %d", netdev->ifindex); + + netdev_enter_ready(netdev); + + return 0; +} + +#define HASH_KEY SD_ID128_MAKE(52,e1,45,bd,00,6f,29,96,21,c6,30,6d,83,71,04,48) + +int netdev_get_mac(const char *ifname, struct ether_addr **ret) { + _cleanup_free_ struct ether_addr *mac = NULL; + uint64_t result; + size_t l, sz; + uint8_t *v; + int r; + + assert(ifname); + assert(ret); + + mac = new0(struct ether_addr, 1); + if (!mac) + return -ENOMEM; + + l = strlen(ifname); + sz = sizeof(sd_id128_t) + l; + v = alloca(sz); + + /* fetch some persistent data unique to the machine */ + r = sd_id128_get_machine((sd_id128_t*) v); + if (r < 0) + return r; + + /* combine with some data unique (on this machine) to this + * netdev */ + memcpy(v + sizeof(sd_id128_t), ifname, l); + + /* Let's hash the host machine ID plus the container name. We + * use a fixed, but originally randomly created hash key here. */ + result = siphash24(v, sz, HASH_KEY.bytes); + + assert_cc(ETH_ALEN <= sizeof(result)); + memcpy(mac->ether_addr_octet, &result, ETH_ALEN); + + /* see eth_random_addr in the kernel */ + mac->ether_addr_octet[0] &= 0xfe; /* clear multicast bit */ + mac->ether_addr_octet[0] |= 0x02; /* set local assignment bit (IEEE802) */ + + *ret = mac; + mac = NULL; + + return 0; +} + +static int netdev_create(NetDev *netdev, Link *link, + sd_netlink_message_handler_t callback) { + int r; + + assert(netdev); + assert(!link || callback); + + /* create netdev */ + if (NETDEV_VTABLE(netdev)->create) { + assert(!link); + + r = NETDEV_VTABLE(netdev)->create(netdev); + if (r < 0) + return r; + + log_netdev_debug(netdev, "Created"); + } else { + _cleanup_(sd_netlink_message_unrefp) sd_netlink_message *m = NULL; + + r = sd_rtnl_message_new_link(netdev->manager->rtnl, &m, RTM_NEWLINK, 0); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not allocate RTM_NEWLINK message: %m"); + + r = sd_netlink_message_append_string(m, IFLA_IFNAME, netdev->ifname); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IFNAME, attribute: %m"); + + if (netdev->mac) { + r = sd_netlink_message_append_ether_addr(m, IFLA_ADDRESS, netdev->mac); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_ADDRESS attribute: %m"); + } + + if (netdev->mtu) { + r = sd_netlink_message_append_u32(m, IFLA_MTU, netdev->mtu); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_MTU attribute: %m"); + } + + if (link) { + r = sd_netlink_message_append_u32(m, IFLA_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_LINK attribute: %m"); + } + + r = sd_netlink_message_open_container(m, IFLA_LINKINFO); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_LINKINFO attribute: %m"); + + r = sd_netlink_message_open_container_union(m, IFLA_INFO_DATA, netdev_kind_to_string(netdev->kind)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_INFO_DATA attribute: %m"); + + if (NETDEV_VTABLE(netdev)->fill_message_create) { + r = NETDEV_VTABLE(netdev)->fill_message_create(netdev, link, m); + if (r < 0) + return r; + } + + r = sd_netlink_message_close_container(m); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_INFO_DATA attribute: %m"); + + r = sd_netlink_message_close_container(m); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_LINKINFO attribute: %m"); + + if (link) { + r = sd_netlink_call_async(netdev->manager->rtnl, m, callback, link, 0, NULL); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not send rtnetlink message: %m"); + + link_ref(link); + } else { + r = sd_netlink_call_async(netdev->manager->rtnl, m, netdev_create_handler, netdev, 0, NULL); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not send rtnetlink message: %m"); + + netdev_ref(netdev); + } + + netdev->state = NETDEV_STATE_CREATING; + + log_netdev_debug(netdev, "Creating"); + } + + return 0; +} + +/* the callback must be called, possibly after a timeout, as otherwise the Link will hang */ +int netdev_join(NetDev *netdev, Link *link, sd_netlink_message_handler_t callback) { + int r; + + assert(netdev); + assert(netdev->manager); + assert(netdev->manager->rtnl); + assert(NETDEV_VTABLE(netdev)); + + switch (NETDEV_VTABLE(netdev)->create_type) { + case NETDEV_CREATE_MASTER: + r = netdev_enslave(netdev, link, callback); + if (r < 0) + return r; + + break; + case NETDEV_CREATE_STACKED: + r = netdev_create(netdev, link, callback); + if (r < 0) + return r; + + break; + default: + assert_not_reached("Can not join independent netdev"); + } + + return 0; +} + +static int netdev_load_one(Manager *manager, const char *filename) { + _cleanup_netdev_unref_ NetDev *netdev = NULL; + _cleanup_free_ NetDev *netdev_raw = NULL; + _cleanup_fclose_ FILE *file = NULL; + const char *dropin_dirname; + int r; + + assert(manager); + assert(filename); + + file = fopen(filename, "re"); + if (!file) { + if (errno == ENOENT) + return 0; + else + return -errno; + } + + if (null_or_empty_fd(fileno(file))) { + log_debug("Skipping empty file: %s", filename); + return 0; + } + + netdev_raw = new0(NetDev, 1); + if (!netdev_raw) + return log_oom(); + + netdev_raw->kind = _NETDEV_KIND_INVALID; + dropin_dirname = strjoina(basename(filename), ".d"); + + r = config_parse_many(filename, network_dirs, dropin_dirname, + "Match\0NetDev\0", + config_item_perf_lookup, network_netdev_gperf_lookup, + true, netdev_raw); + if (r < 0) + return r; + + r = fseek(file, 0, SEEK_SET); + if (r < 0) + return -errno; + + /* skip out early if configuration does not match the environment */ + if (net_match_config(NULL, NULL, NULL, NULL, NULL, + netdev_raw->match_host, netdev_raw->match_virt, + netdev_raw->match_kernel, netdev_raw->match_arch, + NULL, NULL, NULL, NULL, NULL, NULL) <= 0) + return 0; + + if (netdev_raw->kind == _NETDEV_KIND_INVALID) { + log_warning("NetDev has no Kind configured in %s. Ignoring", filename); + return 0; + } + + if (!netdev_raw->ifname) { + log_warning("NetDev without Name configured in %s. Ignoring", filename); + return 0; + } + + netdev = malloc0(NETDEV_VTABLE(netdev_raw)->object_size); + if (!netdev) + return log_oom(); + + netdev->n_ref = 1; + netdev->manager = manager; + netdev->state = _NETDEV_STATE_INVALID; + netdev->kind = netdev_raw->kind; + netdev->ifname = netdev_raw->ifname; + + if (NETDEV_VTABLE(netdev)->init) + NETDEV_VTABLE(netdev)->init(netdev); + + r = config_parse(NULL, filename, file, + NETDEV_VTABLE(netdev)->sections, + config_item_perf_lookup, network_netdev_gperf_lookup, + false, false, false, netdev); + if (r < 0) + return r; + + /* verify configuration */ + if (NETDEV_VTABLE(netdev)->config_verify) { + r = NETDEV_VTABLE(netdev)->config_verify(netdev, filename); + if (r < 0) + return 0; + } + + netdev->filename = strdup(filename); + if (!netdev->filename) + return log_oom(); + + if (!netdev->mac && netdev->kind != NETDEV_KIND_VLAN) { + r = netdev_get_mac(netdev->ifname, &netdev->mac); + if (r < 0) + return log_error_errno(r, "Failed to generate predictable MAC address for %s: %m", netdev->ifname); + } + + r = hashmap_put(netdev->manager->netdevs, netdev->ifname, netdev); + if (r < 0) + return r; + + LIST_HEAD_INIT(netdev->callbacks); + + log_netdev_debug(netdev, "loaded %s", netdev_kind_to_string(netdev->kind)); + + switch (NETDEV_VTABLE(netdev)->create_type) { + case NETDEV_CREATE_MASTER: + case NETDEV_CREATE_INDEPENDENT: + r = netdev_create(netdev, NULL, NULL); + if (r < 0) + return 0; + + break; + default: + break; + } + + netdev = NULL; + + return 0; +} + +int netdev_load(Manager *manager) { + _cleanup_strv_free_ char **files = NULL; + NetDev *netdev; + char **f; + int r; + + assert(manager); + + while ((netdev = hashmap_first(manager->netdevs))) + netdev_unref(netdev); + + r = conf_files_list_strv(&files, ".netdev", NULL, network_dirs); + if (r < 0) + return log_error_errno(r, "Failed to enumerate netdev files: %m"); + + STRV_FOREACH_BACKWARDS(f, files) { + r = netdev_load_one(manager, *f); + if (r < 0) + return r; + } + + return 0; +} diff --git a/src/network/netdev/netdev.h b/src/network/netdev/netdev.h new file mode 100644 index 0000000000..70ff947b99 --- /dev/null +++ b/src/network/netdev/netdev.h @@ -0,0 +1,202 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2013 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "sd-netlink.h" + +#include "list.h" +#include "time-util.h" + +typedef struct netdev_join_callback netdev_join_callback; +typedef struct Link Link; + +struct netdev_join_callback { + sd_netlink_message_handler_t callback; + Link *link; + + LIST_FIELDS(netdev_join_callback, callbacks); +}; + +typedef enum NetDevKind { + NETDEV_KIND_BRIDGE, + NETDEV_KIND_BOND, + NETDEV_KIND_VLAN, + NETDEV_KIND_MACVLAN, + NETDEV_KIND_MACVTAP, + NETDEV_KIND_IPVLAN, + NETDEV_KIND_VXLAN, + NETDEV_KIND_IPIP, + NETDEV_KIND_GRE, + NETDEV_KIND_GRETAP, + NETDEV_KIND_IP6GRE, + NETDEV_KIND_IP6GRETAP, + NETDEV_KIND_SIT, + NETDEV_KIND_VETH, + NETDEV_KIND_VTI, + NETDEV_KIND_VTI6, + NETDEV_KIND_IP6TNL, + NETDEV_KIND_DUMMY, + NETDEV_KIND_TUN, + NETDEV_KIND_TAP, + NETDEV_KIND_VRF, + NETDEV_KIND_VCAN, + _NETDEV_KIND_MAX, + _NETDEV_KIND_INVALID = -1 +} NetDevKind; + +typedef enum NetDevState { + NETDEV_STATE_FAILED, + NETDEV_STATE_CREATING, + NETDEV_STATE_READY, + NETDEV_STATE_LINGER, + _NETDEV_STATE_MAX, + _NETDEV_STATE_INVALID = -1, +} NetDevState; + +typedef enum NetDevCreateType { + NETDEV_CREATE_INDEPENDENT, + NETDEV_CREATE_MASTER, + NETDEV_CREATE_STACKED, + _NETDEV_CREATE_MAX, + _NETDEV_CREATE_INVALID = -1, +} NetDevCreateType; + +typedef struct Manager Manager; +typedef struct Condition Condition; + +typedef struct NetDev { + Manager *manager; + + int n_ref; + + char *filename; + + Condition *match_host; + Condition *match_virt; + Condition *match_kernel; + Condition *match_arch; + + NetDevState state; + NetDevKind kind; + char *description; + char *ifname; + struct ether_addr *mac; + size_t mtu; + int ifindex; + + LIST_HEAD(netdev_join_callback, callbacks); +} NetDev; + +typedef struct NetDevVTable { + /* How much memory does an object of this unit type need */ + size_t object_size; + + /* Config file sections this netdev kind understands, separated + * by NUL chars */ + const char *sections; + + /* This should reset all type-specific variables. This should + * not allocate memory, and is called with zero-initialized + * data. It should hence only initialize variables that need + * to be set != 0. */ + void (*init)(NetDev *n); + + /* This should free all kind-specific variables. It should be + * idempotent. */ + void (*done)(NetDev *n); + + /* fill in message to create netdev */ + int (*fill_message_create)(NetDev *netdev, Link *link, sd_netlink_message *message); + + /* specifies if netdev is independent, or a master device or a stacked device */ + NetDevCreateType create_type; + + /* create netdev, if not done via rtnl */ + int (*create)(NetDev *netdev); + + /* perform additional configuration after netdev has been createad */ + int (*post_create)(NetDev *netdev, Link *link, sd_netlink_message *message); + + /* verify that compulsory configuration options were specified */ + int (*config_verify)(NetDev *netdev, const char *filename); +} NetDevVTable; + +extern const NetDevVTable * const netdev_vtable[_NETDEV_KIND_MAX]; + +#define NETDEV_VTABLE(n) netdev_vtable[(n)->kind] + +/* For casting a netdev into the various netdev kinds */ +#define DEFINE_NETDEV_CAST(UPPERCASE, MixedCase) \ + static inline MixedCase* UPPERCASE(NetDev *n) { \ + if (_unlikely_(!n || n->kind != NETDEV_KIND_##UPPERCASE)) \ + return NULL; \ + \ + return (MixedCase*) n; \ + } + +/* For casting the various netdev kinds into a netdev */ +#define NETDEV(n) (&(n)->meta) + +int netdev_load(Manager *manager); +void netdev_drop(NetDev *netdev); + +NetDev *netdev_unref(NetDev *netdev); +NetDev *netdev_ref(NetDev *netdev); + +DEFINE_TRIVIAL_CLEANUP_FUNC(NetDev*, netdev_unref); +#define _cleanup_netdev_unref_ _cleanup_(netdev_unrefp) + +int netdev_get(Manager *manager, const char *name, NetDev **ret); +int netdev_set_ifindex(NetDev *netdev, sd_netlink_message *newlink); +int netdev_enslave(NetDev *netdev, Link *link, sd_netlink_message_handler_t callback); +int netdev_get_mac(const char *ifname, struct ether_addr **ret); +int netdev_join(NetDev *netdev, Link *link, sd_netlink_message_handler_t cb); + +const char *netdev_kind_to_string(NetDevKind d) _const_; +NetDevKind netdev_kind_from_string(const char *d) _pure_; + +int config_parse_netdev_kind(const char *unit, const char *filename, unsigned line, const char *section, unsigned section_line, const char *lvalue, int ltype, const char *rvalue, void *data, void *userdata); + +/* gperf */ +const struct ConfigPerfItem* network_netdev_gperf_lookup(const char *key, unsigned length); + +/* Macros which append INTERFACE= to the message */ + +#define log_netdev_full(netdev, level, error, ...) \ + ({ \ + const NetDev *_n = (netdev); \ + _n ? log_object_internal(level, error, __FILE__, __LINE__, __func__, "INTERFACE=", _n->ifname, NULL, NULL, ##__VA_ARGS__) : \ + log_internal(level, error, __FILE__, __LINE__, __func__, ##__VA_ARGS__); \ + }) + +#define log_netdev_debug(netdev, ...) log_netdev_full(netdev, LOG_DEBUG, 0, ##__VA_ARGS__) +#define log_netdev_info(netdev, ...) log_netdev_full(netdev, LOG_INFO, 0, ##__VA_ARGS__) +#define log_netdev_notice(netdev, ...) log_netdev_full(netdev, LOG_NOTICE, 0, ##__VA_ARGS__) +#define log_netdev_warning(netdev, ...) log_netdev_full(netdev, LOG_WARNING, 0, ## __VA_ARGS__) +#define log_netdev_error(netdev, ...) log_netdev_full(netdev, LOG_ERR, 0, ##__VA_ARGS__) + +#define log_netdev_debug_errno(netdev, error, ...) log_netdev_full(netdev, LOG_DEBUG, error, ##__VA_ARGS__) +#define log_netdev_info_errno(netdev, error, ...) log_netdev_full(netdev, LOG_INFO, error, ##__VA_ARGS__) +#define log_netdev_notice_errno(netdev, error, ...) log_netdev_full(netdev, LOG_NOTICE, error, ##__VA_ARGS__) +#define log_netdev_warning_errno(netdev, error, ...) log_netdev_full(netdev, LOG_WARNING, error, ##__VA_ARGS__) +#define log_netdev_error_errno(netdev, error, ...) log_netdev_full(netdev, LOG_ERR, error, ##__VA_ARGS__) + +#define LOG_NETDEV_MESSAGE(netdev, fmt, ...) "MESSAGE=%s: " fmt, (netdev)->ifname, ##__VA_ARGS__ +#define LOG_NETDEV_INTERFACE(netdev) "INTERFACE=%s", (netdev)->ifname diff --git a/src/network/netdev/tunnel.c b/src/network/netdev/tunnel.c new file mode 100644 index 0000000000..b03e770061 --- /dev/null +++ b/src/network/netdev/tunnel.c @@ -0,0 +1,731 @@ +/*** + This file is part of systemd. + + Copyright 2014 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <arpa/inet.h> +#include <net/if.h> +#include <linux/ip.h> +#include <linux/if_tunnel.h> +#include <linux/ip6_tunnel.h> + +#include "sd-netlink.h" + +#include "conf-parser.h" +#include "missing.h" +#include "networkd-link.h" +#include "netdev/tunnel.h" +#include "parse-util.h" +#include "string-table.h" +#include "string-util.h" +#include "util.h" + +#define DEFAULT_TNL_HOP_LIMIT 64 +#define IP6_FLOWINFO_FLOWLABEL htobe32(0x000FFFFF) + +static const char* const ip6tnl_mode_table[_NETDEV_IP6_TNL_MODE_MAX] = { + [NETDEV_IP6_TNL_MODE_IP6IP6] = "ip6ip6", + [NETDEV_IP6_TNL_MODE_IPIP6] = "ipip6", + [NETDEV_IP6_TNL_MODE_ANYIP6] = "any", +}; + +DEFINE_STRING_TABLE_LOOKUP(ip6tnl_mode, Ip6TnlMode); +DEFINE_CONFIG_PARSE_ENUM(config_parse_ip6tnl_mode, ip6tnl_mode, Ip6TnlMode, "Failed to parse ip6 tunnel Mode"); + +static int netdev_ipip_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t = IPIP(netdev); + int r; + + assert(netdev); + assert(link); + assert(m); + assert(t); + assert(IN_SET(t->family, AF_INET, AF_UNSPEC)); + + r = sd_netlink_message_append_u32(m, IFLA_IPTUN_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LINK attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_IPTUN_LOCAL, &t->local.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_IPTUN_REMOTE, &t->remote.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_REMOTE attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_TTL, t->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_TTL attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_PMTUDISC, t->pmtudisc); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_PMTUDISC attribute: %m"); + + return r; +} + +static int netdev_sit_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t = SIT(netdev); + int r; + + assert(netdev); + assert(link); + assert(m); + assert(t); + assert(IN_SET(t->family, AF_INET, AF_UNSPEC)); + + r = sd_netlink_message_append_u32(m, IFLA_IPTUN_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LINK attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_IPTUN_LOCAL, &t->local.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_IPTUN_REMOTE, &t->remote.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_REMOTE attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_TTL, t->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_TTL attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_PMTUDISC, t->pmtudisc); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_PMTUDISC attribute: %m"); + + return r; +} + +static int netdev_gre_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t; + int r; + + assert(netdev); + + if (netdev->kind == NETDEV_KIND_GRE) + t = GRE(netdev); + else + t = GRETAP(netdev); + + assert(t); + assert(IN_SET(t->family, AF_INET, AF_UNSPEC)); + assert(link); + assert(m); + + r = sd_netlink_message_append_u32(m, IFLA_GRE_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_LINK attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_GRE_LOCAL, &t->local.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_GRE_REMOTE, &t->remote.in); + if (r < 0) + log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_REMOTE attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_GRE_TTL, t->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_TTL attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_GRE_TOS, t->tos); + if (r < 0) + log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_TOS attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_GRE_PMTUDISC, t->pmtudisc); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_PMTUDISC attribute: %m"); + + return r; +} + +static int netdev_ip6gre_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t; + int r; + + assert(netdev); + + if (netdev->kind == NETDEV_KIND_IP6GRE) + t = IP6GRE(netdev); + else + t = IP6GRETAP(netdev); + + assert(t); + assert(t->family == AF_INET6); + assert(link); + assert(m); + + r = sd_netlink_message_append_u32(m, IFLA_GRE_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_LINK attribute: %m"); + + r = sd_netlink_message_append_in6_addr(m, IFLA_GRE_LOCAL, &t->local.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in6_addr(m, IFLA_GRE_REMOTE, &t->remote.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_REMOTE attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_GRE_TTL, t->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_TTL attribute: %m"); + + if (t->ipv6_flowlabel != _NETDEV_IPV6_FLOWLABEL_INVALID) { + r = sd_netlink_message_append_u32(m, IFLA_GRE_FLOWINFO, t->ipv6_flowlabel); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_FLOWINFO attribute: %m"); + } + + r = sd_netlink_message_append_u32(m, IFLA_GRE_FLAGS, t->flags); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_GRE_FLAGS attribute: %m"); + + return r; +} + +static int netdev_vti_fill_message_key(NetDev *netdev, Link *link, sd_netlink_message *m) { + uint32_t ikey, okey; + Tunnel *t; + int r; + + assert(link); + assert(m); + + if (netdev->kind == NETDEV_KIND_VTI) + t = VTI(netdev); + else + t = VTI6(netdev); + + assert(t); + + if (t->key != 0) + ikey = okey = htobe32(t->key); + else { + ikey = htobe32(t->ikey); + okey = htobe32(t->okey); + } + + r = sd_netlink_message_append_u32(m, IFLA_VTI_IKEY, ikey); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VTI_IKEY attribute: %m"); + + r = sd_netlink_message_append_u32(m, IFLA_VTI_OKEY, okey); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VTI_OKEY attribute: %m"); + + return 0; +} + +static int netdev_vti_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t = VTI(netdev); + int r; + + assert(netdev); + assert(link); + assert(m); + assert(t); + assert(t->family == AF_INET); + + r = sd_netlink_message_append_u32(m, IFLA_VTI_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LINK attribute: %m"); + + r = netdev_vti_fill_message_key(netdev, link, m); + if (r < 0) + return r; + + r = sd_netlink_message_append_in_addr(m, IFLA_VTI_LOCAL, &t->local.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in_addr(m, IFLA_VTI_REMOTE, &t->remote.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_REMOTE attribute: %m"); + + return r; +} + +static int netdev_vti6_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t = VTI6(netdev); + int r; + + assert(netdev); + assert(link); + assert(m); + assert(t); + assert(t->family == AF_INET6); + + r = sd_netlink_message_append_u32(m, IFLA_VTI_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LINK attribute: %m"); + + r = netdev_vti_fill_message_key(netdev, link, m); + if (r < 0) + return r; + + r = sd_netlink_message_append_in6_addr(m, IFLA_VTI_LOCAL, &t->local.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in6_addr(m, IFLA_VTI_REMOTE, &t->remote.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_REMOTE attribute: %m"); + + return r; +} + +static int netdev_ip6tnl_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Tunnel *t = IP6TNL(netdev); + uint8_t proto; + int r; + + assert(netdev); + assert(link); + assert(m); + assert(t); + assert(t->family == AF_INET6); + + r = sd_netlink_message_append_u32(m, IFLA_IPTUN_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LINK attribute: %m"); + + r = sd_netlink_message_append_in6_addr(m, IFLA_IPTUN_LOCAL, &t->local.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_LOCAL attribute: %m"); + + r = sd_netlink_message_append_in6_addr(m, IFLA_IPTUN_REMOTE, &t->remote.in6); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_REMOTE attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_TTL, t->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_TTL attribute: %m"); + + if (t->ipv6_flowlabel != _NETDEV_IPV6_FLOWLABEL_INVALID) { + r = sd_netlink_message_append_u32(m, IFLA_IPTUN_FLOWINFO, t->ipv6_flowlabel); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_FLOWINFO attribute: %m"); + } + + if (t->copy_dscp) + t->flags |= IP6_TNL_F_RCV_DSCP_COPY; + + if (t->encap_limit != IPV6_DEFAULT_TNL_ENCAP_LIMIT) { + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_ENCAP_LIMIT, t->encap_limit); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_ENCAP_LIMIT attribute: %m"); + } + + r = sd_netlink_message_append_u32(m, IFLA_IPTUN_FLAGS, t->flags); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_FLAGS attribute: %m"); + + switch (t->ip6tnl_mode) { + case NETDEV_IP6_TNL_MODE_IP6IP6: + proto = IPPROTO_IPV6; + break; + case NETDEV_IP6_TNL_MODE_IPIP6: + proto = IPPROTO_IPIP; + break; + case NETDEV_IP6_TNL_MODE_ANYIP6: + default: + proto = 0; + break; + } + + r = sd_netlink_message_append_u8(m, IFLA_IPTUN_PROTO, proto); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_IPTUN_MODE attribute: %m"); + + return r; +} + +static int netdev_tunnel_verify(NetDev *netdev, const char *filename) { + Tunnel *t = NULL; + + assert(netdev); + assert(filename); + + switch (netdev->kind) { + case NETDEV_KIND_IPIP: + t = IPIP(netdev); + break; + case NETDEV_KIND_SIT: + t = SIT(netdev); + break; + case NETDEV_KIND_GRE: + t = GRE(netdev); + break; + case NETDEV_KIND_GRETAP: + t = GRETAP(netdev); + break; + case NETDEV_KIND_IP6GRE: + t = IP6GRE(netdev); + break; + case NETDEV_KIND_IP6GRETAP: + t = IP6GRETAP(netdev); + break; + case NETDEV_KIND_VTI: + t = VTI(netdev); + break; + case NETDEV_KIND_VTI6: + t = VTI6(netdev); + break; + case NETDEV_KIND_IP6TNL: + t = IP6TNL(netdev); + break; + default: + assert_not_reached("Invalid tunnel kind"); + } + + assert(t); + + if (t->family != AF_INET && t->family != AF_INET6 && t->family != 0) { + log_warning("Tunnel with invalid address family configured in %s. Ignoring", filename); + return -EINVAL; + } + + if (netdev->kind == NETDEV_KIND_IP6TNL) { + if (t->ip6tnl_mode == _NETDEV_IP6_TNL_MODE_INVALID) { + log_warning("IP6 Tunnel without mode configured in %s. Ignoring", filename); + return -EINVAL; + } + } + + return 0; +} + +int config_parse_tunnel_address(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + Tunnel *t = userdata; + union in_addr_union *addr = data, buffer; + int r, f; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + if (streq(rvalue, "any")) { + t->family = 0; + return 0; + } else { + + r = in_addr_from_string_auto(rvalue, &f, &buffer); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Tunnel address is invalid, ignoring assignment: %s", rvalue); + return 0; + } + + if (t->family != AF_UNSPEC && t->family != f) { + log_syntax(unit, LOG_ERR, filename, line, 0, "Tunnel addresses incompatible, ignoring assignment: %s", rvalue); + return 0; + } + } + + t->family = f; + *addr = buffer; + + return 0; +} + +int config_parse_tunnel_key(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + union in_addr_union buffer; + Tunnel *t = userdata; + uint32_t k; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + r = in_addr_from_string(AF_INET, rvalue, &buffer); + if (r < 0) { + r = safe_atou32(rvalue, &k); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, 0, "Failed to parse tunnel key ignoring assignment: %s", rvalue); + return 0; + } + } else + k = be32toh(buffer.in.s_addr); + + if (streq(lvalue, "Key")) + t->key = k; + else if (streq(lvalue, "InputKey")) + t->ikey = k; + else + t->okey = k; + + return 0; +} + +int config_parse_ipv6_flowlabel(const char* unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + IPv6FlowLabel *ipv6_flowlabel = data; + Tunnel *t = userdata; + int k = 0; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(ipv6_flowlabel); + + if (streq(rvalue, "inherit")) { + *ipv6_flowlabel = IP6_FLOWINFO_FLOWLABEL; + t->flags |= IP6_TNL_F_USE_ORIG_FLOWLABEL; + } else { + r = config_parse_int(unit, filename, line, section, section_line, lvalue, ltype, rvalue, &k, userdata); + if (r < 0) + return r; + + if (k > 0xFFFFF) + log_syntax(unit, LOG_ERR, filename, line, 0, "Failed to parse IPv6 flowlabel option, ignoring: %s", rvalue); + else { + *ipv6_flowlabel = htobe32(k) & IP6_FLOWINFO_FLOWLABEL; + t->flags &= ~IP6_TNL_F_USE_ORIG_FLOWLABEL; + } + } + + return 0; +} + +int config_parse_encap_limit(const char* unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + Tunnel *t = userdata; + int k = 0; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + + if (streq(rvalue, "none")) + t->flags |= IP6_TNL_F_IGN_ENCAP_LIMIT; + else { + r = safe_atoi(rvalue, &k); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse Tunnel Encapsulation Limit option, ignoring: %s", rvalue); + return 0; + } + + if (k > 255 || k < 0) + log_syntax(unit, LOG_ERR, filename, line, 0, "Invalid Tunnel Encapsulation value, ignoring: %d", k); + else { + t->encap_limit = k; + t->flags &= ~IP6_TNL_F_IGN_ENCAP_LIMIT; + } + } + + return 0; +} + +static void ipip_init(NetDev *n) { + Tunnel *t = IPIP(n); + + assert(n); + assert(t); + + t->pmtudisc = true; + t->family = AF_UNSPEC; +} + +static void sit_init(NetDev *n) { + Tunnel *t = SIT(n); + + assert(n); + assert(t); + + t->pmtudisc = true; + t->family = AF_UNSPEC; +} + +static void vti_init(NetDev *n) { + Tunnel *t; + + assert(n); + + if (n->kind == NETDEV_KIND_VTI) + t = VTI(n); + else + t = VTI6(n); + + assert(t); + + t->pmtudisc = true; +} + +static void gre_init(NetDev *n) { + Tunnel *t; + + assert(n); + + if (n->kind == NETDEV_KIND_GRE) + t = GRE(n); + else + t = GRETAP(n); + + assert(t); + + t->pmtudisc = true; + t->family = AF_UNSPEC; +} + +static void ip6gre_init(NetDev *n) { + Tunnel *t; + + assert(n); + + if (n->kind == NETDEV_KIND_IP6GRE) + t = IP6GRE(n); + else + t = IP6GRETAP(n); + + assert(t); + + t->ttl = DEFAULT_TNL_HOP_LIMIT; +} + +static void ip6tnl_init(NetDev *n) { + Tunnel *t = IP6TNL(n); + + assert(n); + assert(t); + + t->ttl = DEFAULT_TNL_HOP_LIMIT; + t->encap_limit = IPV6_DEFAULT_TNL_ENCAP_LIMIT; + t->ip6tnl_mode = _NETDEV_IP6_TNL_MODE_INVALID; + t->ipv6_flowlabel = _NETDEV_IPV6_FLOWLABEL_INVALID; +} + +const NetDevVTable ipip_vtable = { + .object_size = sizeof(Tunnel), + .init = ipip_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_ipip_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable sit_vtable = { + .object_size = sizeof(Tunnel), + .init = sit_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_sit_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable vti_vtable = { + .object_size = sizeof(Tunnel), + .init = vti_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_vti_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable vti6_vtable = { + .object_size = sizeof(Tunnel), + .init = vti_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_vti6_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable gre_vtable = { + .object_size = sizeof(Tunnel), + .init = gre_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_gre_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable gretap_vtable = { + .object_size = sizeof(Tunnel), + .init = gre_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_gre_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable ip6gre_vtable = { + .object_size = sizeof(Tunnel), + .init = ip6gre_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_ip6gre_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable ip6gretap_vtable = { + .object_size = sizeof(Tunnel), + .init = ip6gre_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_ip6gre_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; + +const NetDevVTable ip6tnl_vtable = { + .object_size = sizeof(Tunnel), + .init = ip6tnl_init, + .sections = "Match\0NetDev\0Tunnel\0", + .fill_message_create = netdev_ip6tnl_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_tunnel_verify, +}; diff --git a/src/network/netdev/tunnel.h b/src/network/netdev/tunnel.h new file mode 100644 index 0000000000..d78c6135ee --- /dev/null +++ b/src/network/netdev/tunnel.h @@ -0,0 +1,119 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "in-addr-util.h" + +#include "netdev/netdev.h" + +typedef enum Ip6TnlMode { + NETDEV_IP6_TNL_MODE_IP6IP6, + NETDEV_IP6_TNL_MODE_IPIP6, + NETDEV_IP6_TNL_MODE_ANYIP6, + _NETDEV_IP6_TNL_MODE_MAX, + _NETDEV_IP6_TNL_MODE_INVALID = -1, +} Ip6TnlMode; + +typedef enum IPv6FlowLabel { + NETDEV_IPV6_FLOWLABEL_INHERIT = 0xFFFFF + 1, + _NETDEV_IPV6_FLOWLABEL_MAX, + _NETDEV_IPV6_FLOWLABEL_INVALID = -1, +} IPv6FlowLabel; + +typedef struct Tunnel { + NetDev meta; + + uint8_t encap_limit; + + int family; + int ipv6_flowlabel; + + unsigned ttl; + unsigned tos; + unsigned flags; + + uint32_t key; + uint32_t ikey; + uint32_t okey; + + union in_addr_union local; + union in_addr_union remote; + + Ip6TnlMode ip6tnl_mode; + + bool pmtudisc; + bool copy_dscp; +} Tunnel; + +DEFINE_NETDEV_CAST(IPIP, Tunnel); +DEFINE_NETDEV_CAST(GRE, Tunnel); +DEFINE_NETDEV_CAST(GRETAP, Tunnel); +DEFINE_NETDEV_CAST(IP6GRE, Tunnel); +DEFINE_NETDEV_CAST(IP6GRETAP, Tunnel); +DEFINE_NETDEV_CAST(SIT, Tunnel); +DEFINE_NETDEV_CAST(VTI, Tunnel); +DEFINE_NETDEV_CAST(VTI6, Tunnel); +DEFINE_NETDEV_CAST(IP6TNL, Tunnel); +extern const NetDevVTable ipip_vtable; +extern const NetDevVTable sit_vtable; +extern const NetDevVTable vti_vtable; +extern const NetDevVTable vti6_vtable; +extern const NetDevVTable gre_vtable; +extern const NetDevVTable gretap_vtable; +extern const NetDevVTable ip6gre_vtable; +extern const NetDevVTable ip6gretap_vtable; +extern const NetDevVTable ip6tnl_vtable; + +const char *ip6tnl_mode_to_string(Ip6TnlMode d) _const_; +Ip6TnlMode ip6tnl_mode_from_string(const char *d) _pure_; + +int config_parse_ip6tnl_mode(const char *unit, const char *filename, + unsigned line, const char *section, + unsigned section_line, const char *lvalue, + int ltype, const char *rvalue, void *data, + void *userdata); + +int config_parse_tunnel_address(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata); + +int config_parse_ipv6_flowlabel(const char *unit, const char *filename, + unsigned line, const char *section, + unsigned section_line, const char *lvalue, + int ltype, const char *rvalue, void *data, + void *userdata); + +int config_parse_encap_limit(const char *unit, const char *filename, + unsigned line, const char *section, + unsigned section_line, const char *lvalue, + int ltype, const char *rvalue, void *data, + void *userdata); +int config_parse_tunnel_key(const char *unit, const char *filename, + unsigned line, const char *section, + unsigned section_line, const char *lvalue, + int ltype, const char *rvalue, void *data, + void *userdata); diff --git a/src/network/netdev/tuntap.c b/src/network/netdev/tuntap.c new file mode 100644 index 0000000000..3d62808842 --- /dev/null +++ b/src/network/netdev/tuntap.c @@ -0,0 +1,183 @@ +/*** + This file is part of systemd. + + Copyright 2014 Susant Sahani <susant@redhat.com> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <fcntl.h> +#include <linux/if_tun.h> +#include <net/if.h> +#include <netinet/if_ether.h> +#include <sys/ioctl.h> +#include <sys/stat.h> +#include <sys/types.h> + +#include "alloc-util.h" +#include "fd-util.h" +#include "netdev/tuntap.h" +#include "user-util.h" + +#define TUN_DEV "/dev/net/tun" + +static int netdev_fill_tuntap_message(NetDev *netdev, struct ifreq *ifr) { + TunTap *t; + + assert(netdev); + assert(netdev->ifname); + assert(ifr); + + if (netdev->kind == NETDEV_KIND_TAP) { + t = TAP(netdev); + ifr->ifr_flags |= IFF_TAP; + } else { + t = TUN(netdev); + ifr->ifr_flags |= IFF_TUN; + } + + if (!t->packet_info) + ifr->ifr_flags |= IFF_NO_PI; + + if (t->one_queue) + ifr->ifr_flags |= IFF_ONE_QUEUE; + + if (t->multi_queue) + ifr->ifr_flags |= IFF_MULTI_QUEUE; + + if (t->vnet_hdr) + ifr->ifr_flags |= IFF_VNET_HDR; + + strncpy(ifr->ifr_name, netdev->ifname, IFNAMSIZ-1); + + return 0; +} + +static int netdev_tuntap_add(NetDev *netdev, struct ifreq *ifr) { + _cleanup_close_ int fd; + TunTap *t = NULL; + const char *user; + const char *group; + uid_t uid; + gid_t gid; + int r; + + assert(netdev); + assert(ifr); + + fd = open(TUN_DEV, O_RDWR); + if (fd < 0) + return log_netdev_error_errno(netdev, -errno, "Failed to open tun dev: %m"); + + r = ioctl(fd, TUNSETIFF, ifr); + if (r < 0) + return log_netdev_error_errno(netdev, -errno, "TUNSETIFF failed on tun dev: %m"); + + if (netdev->kind == NETDEV_KIND_TAP) + t = TAP(netdev); + else + t = TUN(netdev); + + assert(t); + + if (t->user_name) { + + user = t->user_name; + + r = get_user_creds(&user, &uid, NULL, NULL, NULL); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Cannot resolve user name %s: %m", t->user_name); + + r = ioctl(fd, TUNSETOWNER, uid); + if (r < 0) + return log_netdev_error_errno(netdev, -errno, "TUNSETOWNER failed on tun dev: %m"); + } + + if (t->group_name) { + + group = t->group_name; + + r = get_group_creds(&group, &gid); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Cannot resolve group name %s: %m", t->group_name); + + r = ioctl(fd, TUNSETGROUP, gid); + if (r < 0) + return log_netdev_error_errno(netdev, -errno, "TUNSETGROUP failed on tun dev: %m"); + + } + + r = ioctl(fd, TUNSETPERSIST, 1); + if (r < 0) + return log_netdev_error_errno(netdev, -errno, "TUNSETPERSIST failed on tun dev: %m"); + + return 0; +} + +static int netdev_create_tuntap(NetDev *netdev) { + struct ifreq ifr = {}; + int r; + + r = netdev_fill_tuntap_message(netdev, &ifr); + if (r < 0) + return r; + + return netdev_tuntap_add(netdev, &ifr); +} + +static void tuntap_done(NetDev *netdev) { + TunTap *t = NULL; + + assert(netdev); + + if (netdev->kind == NETDEV_KIND_TUN) + t = TUN(netdev); + else + t = TAP(netdev); + + assert(t); + + t->user_name = mfree(t->user_name); + t->group_name = mfree(t->group_name); +} + +static int tuntap_verify(NetDev *netdev, const char *filename) { + assert(netdev); + + if (netdev->mtu) + log_netdev_warning(netdev, "MTU configured for %s, ignoring", netdev_kind_to_string(netdev->kind)); + + if (netdev->mac) + log_netdev_warning(netdev, "MAC configured for %s, ignoring", netdev_kind_to_string(netdev->kind)); + + return 0; +} + +const NetDevVTable tun_vtable = { + .object_size = sizeof(TunTap), + .sections = "Match\0NetDev\0Tun\0", + .config_verify = tuntap_verify, + .done = tuntap_done, + .create = netdev_create_tuntap, + .create_type = NETDEV_CREATE_INDEPENDENT, +}; + +const NetDevVTable tap_vtable = { + .object_size = sizeof(TunTap), + .sections = "Match\0NetDev\0Tap\0", + .config_verify = tuntap_verify, + .done = tuntap_done, + .create = netdev_create_tuntap, + .create_type = NETDEV_CREATE_INDEPENDENT, +}; diff --git a/src/network/netdev/tuntap.h b/src/network/netdev/tuntap.h new file mode 100644 index 0000000000..95d3fcf1e9 --- /dev/null +++ b/src/network/netdev/tuntap.h @@ -0,0 +1,40 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct TunTap TunTap; + +#include "netdev/netdev.h" + +struct TunTap { + NetDev meta; + + char *user_name; + char *group_name; + bool one_queue; + bool multi_queue; + bool packet_info; + bool vnet_hdr; +}; + +DEFINE_NETDEV_CAST(TUN, TunTap); +DEFINE_NETDEV_CAST(TAP, TunTap); +extern const NetDevVTable tun_vtable; +extern const NetDevVTable tap_vtable; diff --git a/src/network/netdev/vcan.c b/src/network/netdev/vcan.c new file mode 100644 index 0000000000..7f56702938 --- /dev/null +++ b/src/network/netdev/vcan.c @@ -0,0 +1,25 @@ +/*** + This file is part of systemd. + + Copyright 2016 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include "netdev/vcan.h" + +const NetDevVTable vcan_vtable = { + .object_size = sizeof(VCan), + .create_type = NETDEV_CREATE_INDEPENDENT, +}; diff --git a/src/network/netdev/vcan.h b/src/network/netdev/vcan.h new file mode 100644 index 0000000000..00838b7675 --- /dev/null +++ b/src/network/netdev/vcan.h @@ -0,0 +1,34 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2016 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct VCan VCan; + +#include <linux/can/netlink.h> + +#include "netdev/netdev.h" + +struct VCan { + NetDev meta; +}; + +DEFINE_NETDEV_CAST(VCAN, VCan); + +extern const NetDevVTable vcan_vtable; diff --git a/src/network/netdev/veth.c b/src/network/netdev/veth.c new file mode 100644 index 0000000000..350b59bf03 --- /dev/null +++ b/src/network/netdev/veth.c @@ -0,0 +1,111 @@ +/*** + This file is part of systemd. + + Copyright 2014 Susant Sahani <susant@redhat.com> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> +#include <linux/veth.h> + +#include "sd-netlink.h" + +#include "netdev/veth.h" + +static int netdev_veth_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Veth *v; + int r; + + assert(netdev); + assert(!link); + assert(m); + + v = VETH(netdev); + + assert(v); + + r = sd_netlink_message_open_container(m, VETH_INFO_PEER); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append VETH_INFO_PEER attribute: %m"); + + if (v->ifname_peer) { + r = sd_netlink_message_append_string(m, IFLA_IFNAME, v->ifname_peer); + if (r < 0) + return log_error_errno(r, "Failed to add netlink interface name: %m"); + } + + if (v->mac_peer) { + r = sd_netlink_message_append_ether_addr(m, IFLA_ADDRESS, v->mac_peer); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_ADDRESS attribute: %m"); + } + + r = sd_netlink_message_close_container(m); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_INFO_DATA attribute: %m"); + + return r; +} + +static int netdev_veth_verify(NetDev *netdev, const char *filename) { + Veth *v; + int r; + + assert(netdev); + assert(filename); + + v = VETH(netdev); + + assert(v); + + if (!v->ifname_peer) { + log_warning("Veth NetDev without peer name configured in %s. Ignoring", + filename); + return -EINVAL; + } + + if (!v->mac_peer) { + r = netdev_get_mac(v->ifname_peer, &v->mac_peer); + if (r < 0) { + log_warning("Failed to generate predictable MAC address for %s. Ignoring", + v->ifname_peer); + return -EINVAL; + } + } + + return 0; +} + +static void veth_done(NetDev *n) { + Veth *v; + + assert(n); + + v = VETH(n); + + assert(v); + + free(v->ifname_peer); + free(v->mac_peer); +} + +const NetDevVTable veth_vtable = { + .object_size = sizeof(Veth), + .sections = "Match\0NetDev\0Peer\0", + .done = veth_done, + .fill_message_create = netdev_veth_fill_message_create, + .create_type = NETDEV_CREATE_INDEPENDENT, + .config_verify = netdev_veth_verify, +}; diff --git a/src/network/netdev/veth.h b/src/network/netdev/veth.h new file mode 100644 index 0000000000..b00ce476e8 --- /dev/null +++ b/src/network/netdev/veth.h @@ -0,0 +1,34 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct Veth Veth; + +#include "netdev/netdev.h" + +struct Veth { + NetDev meta; + + char *ifname_peer; + struct ether_addr *mac_peer; +}; + +DEFINE_NETDEV_CAST(VETH, Veth); +extern const NetDevVTable veth_vtable; diff --git a/src/network/netdev/vlan.c b/src/network/netdev/vlan.c new file mode 100644 index 0000000000..28c061fa4f --- /dev/null +++ b/src/network/netdev/vlan.c @@ -0,0 +1,78 @@ +/*** + This file is part of systemd. + + Copyright 2013 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "netdev/vlan.h" +#include "vlan-util.h" + +static int netdev_vlan_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *req) { + VLan *v; + int r; + + assert(netdev); + assert(link); + assert(req); + + v = VLAN(netdev); + + assert(v); + + r = sd_netlink_message_append_u16(req, IFLA_VLAN_ID, v->id); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VLAN_ID attribute: %m"); + + return 0; +} + +static int netdev_vlan_verify(NetDev *netdev, const char *filename) { + VLan *v; + + assert(netdev); + assert(filename); + + v = VLAN(netdev); + + assert(v); + + if (v->id == VLANID_INVALID) { + log_warning("VLAN without valid Id (%"PRIu16") configured in %s.", v->id, filename); + return -EINVAL; + } + + return 0; +} + +static void vlan_init(NetDev *netdev) { + VLan *v = VLAN(netdev); + + assert(netdev); + assert(v); + + v->id = VLANID_INVALID; +} + +const NetDevVTable vlan_vtable = { + .object_size = sizeof(VLan), + .init = vlan_init, + .sections = "Match\0NetDev\0VLAN\0", + .fill_message_create = netdev_vlan_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_vlan_verify, +}; diff --git a/src/network/netdev/vlan.h b/src/network/netdev/vlan.h new file mode 100644 index 0000000000..fade899997 --- /dev/null +++ b/src/network/netdev/vlan.h @@ -0,0 +1,33 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct VLan VLan; + +#include "netdev/netdev.h" + +struct VLan { + NetDev meta; + + uint16_t id; +}; + +DEFINE_NETDEV_CAST(VLAN, VLan); +extern const NetDevVTable vlan_vtable; diff --git a/src/network/netdev/vrf.c b/src/network/netdev/vrf.c new file mode 100644 index 0000000000..f48b413102 --- /dev/null +++ b/src/network/netdev/vrf.c @@ -0,0 +1,50 @@ +/*** + This file is part of systemd. + + Copyright 2016 Andreas Rammhold <andreas@rammhold.de> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "sd-netlink.h" +#include "missing.h" +#include "netdev/vrf.h" + +static int netdev_vrf_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + Vrf *v; + int r; + + assert(netdev); + assert(!link); + assert(m); + + v = VRF(netdev); + + assert(v); + + r = sd_netlink_message_append_u32(m, IFLA_VRF_TABLE, v->table_id); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IPLA_VRF_TABLE attribute: %m"); + + return r; +} + +const NetDevVTable vrf_vtable = { + .object_size = sizeof(Vrf), + .sections = "NetDev\0VRF\0", + .fill_message_create = netdev_vrf_fill_message_create, + .create_type = NETDEV_CREATE_MASTER, +}; diff --git a/src/network/netdev/vrf.h b/src/network/netdev/vrf.h new file mode 100644 index 0000000000..00f54ed96d --- /dev/null +++ b/src/network/netdev/vrf.h @@ -0,0 +1,33 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2016 Andreas Rammhold <andreas@rammhold.de> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct Vrf Vrf; + +#include "netdev/netdev.h" + +struct Vrf { + NetDev meta; + + uint32_t table_id; +}; + +DEFINE_NETDEV_CAST(VRF, Vrf); +extern const NetDevVTable vrf_vtable; diff --git a/src/network/netdev/vxlan.c b/src/network/netdev/vxlan.c new file mode 100644 index 0000000000..10c892b044 --- /dev/null +++ b/src/network/netdev/vxlan.c @@ -0,0 +1,304 @@ +/*** + This file is part of systemd. + + Copyright 2014 Susant Sahani + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +#include <net/if.h> + +#include "sd-netlink.h" + +#include "conf-parser.h" +#include "alloc-util.h" +#include "extract-word.h" +#include "parse-util.h" +#include "missing.h" + +#include "networkd-link.h" +#include "netdev/vxlan.h" + +static int netdev_vxlan_fill_message_create(NetDev *netdev, Link *link, sd_netlink_message *m) { + VxLan *v; + int r; + + assert(netdev); + assert(link); + assert(m); + + v = VXLAN(netdev); + + assert(v); + + if (v->id <= VXLAN_VID_MAX) { + r = sd_netlink_message_append_u32(m, IFLA_VXLAN_ID, v->id); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_ID attribute: %m"); + } + + r = sd_netlink_message_append_in_addr(m, IFLA_VXLAN_GROUP, &v->group.in); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_GROUP attribute: %m"); + + r = sd_netlink_message_append_u32(m, IFLA_VXLAN_LINK, link->ifindex); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_LINK attribute: %m"); + + if (v->ttl) { + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_TTL, v->ttl); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_TTL attribute: %m"); + } + + if (v->tos) { + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_TOS, v->tos); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_TOS attribute: %m"); + } + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_LEARNING, v->learning); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_LEARNING attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_RSC, v->route_short_circuit); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_RSC attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_PROXY, v->arp_proxy); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_PROXY attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_L2MISS, v->l2miss); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_L2MISS attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_L3MISS, v->l3miss); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_L3MISS attribute: %m"); + + if (v->fdb_ageing) { + r = sd_netlink_message_append_u32(m, IFLA_VXLAN_AGEING, v->fdb_ageing / USEC_PER_SEC); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_AGEING attribute: %m"); + } + + if (v->max_fdb) { + r = sd_netlink_message_append_u32(m, IFLA_VXLAN_LIMIT, v->max_fdb); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_LIMIT attribute: %m"); + } + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_UDP_CSUM, v->udpcsum); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_UDP_CSUM attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_UDP_ZERO_CSUM6_TX, v->udp6zerocsumtx); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_UDP_ZERO_CSUM6_TX attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_UDP_ZERO_CSUM6_RX, v->udp6zerocsumrx); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_UDP_ZERO_CSUM6_RX attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_REMCSUM_TX, v->remote_csum_tx); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_REMCSUM_TX attribute: %m"); + + r = sd_netlink_message_append_u8(m, IFLA_VXLAN_REMCSUM_RX, v->remote_csum_rx); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_REMCSUM_RX attribute: %m"); + + r = sd_netlink_message_append_u16(m, IFLA_VXLAN_PORT, htobe16(v->dest_port)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_PORT attribute: %m"); + + if (v->port_range.low || v->port_range.high) { + struct ifla_vxlan_port_range port_range; + + port_range.low = htobe16(v->port_range.low); + port_range.high = htobe16(v->port_range.high); + + r = sd_netlink_message_append_data(m, IFLA_VXLAN_PORT_RANGE, &port_range, sizeof(port_range)); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_PORT_RANGE attribute: %m"); + } + + if (v->group_policy) { + r = sd_netlink_message_append_flag(m, IFLA_VXLAN_GBP); + if (r < 0) + return log_netdev_error_errno(netdev, r, "Could not append IFLA_VXLAN_GBP attribute: %m"); + } + + return r; +} + +int config_parse_vxlan_group_address(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + VxLan *v = userdata; + union in_addr_union *addr = data, buffer; + int r, f; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + r = in_addr_from_string_auto(rvalue, &f, &buffer); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "vxlan multicast group address is invalid, ignoring assignment: %s", rvalue); + return 0; + } + + if (v->family != AF_UNSPEC && v->family != f) { + log_syntax(unit, LOG_ERR, filename, line, 0, "vxlan multicast group incompatible, ignoring assignment: %s", rvalue); + return 0; + } + + v->family = f; + *addr = buffer; + + return 0; +} + +int config_parse_port_range(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + _cleanup_free_ char *word = NULL; + VxLan *v = userdata; + unsigned low, high; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + r = extract_first_word(&rvalue, &word, NULL, 0); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Failed to extract VXLAN port range, ignoring: %s", rvalue); + return 0; + } + + if (r == 0) + return 0; + + r = parse_range(word, &low, &high); + if (r < 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse VXLAN port range '%s'", word); + return 0; + } + + if (low <= 0 || low > 65535 || high <= 0 || high > 65535) { + log_syntax(unit, LOG_ERR, filename, line, r, + "Failed to parse VXLAN port range '%s'. Port should be greater than 0 and less than 65535.", word); + return 0; + } + + if (high < low) { + log_syntax(unit, LOG_ERR, filename, line, r, + "Failed to parse VXLAN port range '%s'. Port range %u .. %u not valid", word, low, high); + return 0; + } + + v->port_range.low = low; + v->port_range.high = high; + + return 0; +} + +int config_parse_destination_port(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata) { + VxLan *v = userdata; + uint16_t port; + int r; + + assert(filename); + assert(lvalue); + assert(rvalue); + assert(data); + + r = safe_atou16(rvalue, &port); + if (r < 0 || port <= 0) { + log_syntax(unit, LOG_ERR, filename, line, r, "Failed to parse VXLAN destination port '%s'.", rvalue); + return 0; + } + + v->dest_port = port; + + return 0; +} + +static int netdev_vxlan_verify(NetDev *netdev, const char *filename) { + VxLan *v = VXLAN(netdev); + + assert(netdev); + assert(v); + assert(filename); + + if (v->id > VXLAN_VID_MAX) { + log_warning("VXLAN without valid Id configured in %s. Ignoring", filename); + return -EINVAL; + } + + return 0; +} + +static void vxlan_init(NetDev *netdev) { + VxLan *v; + + assert(netdev); + + v = VXLAN(netdev); + + assert(v); + + v->id = VXLAN_VID_MAX + 1; + v->learning = true; + v->udpcsum = false; + v->udp6zerocsumtx = false; + v->udp6zerocsumrx = false; +} + +const NetDevVTable vxlan_vtable = { + .object_size = sizeof(VxLan), + .init = vxlan_init, + .sections = "Match\0NetDev\0VXLAN\0", + .fill_message_create = netdev_vxlan_fill_message_create, + .create_type = NETDEV_CREATE_STACKED, + .config_verify = netdev_vxlan_verify, +}; diff --git a/src/network/netdev/vxlan.h b/src/network/netdev/vxlan.h new file mode 100644 index 0000000000..6c3081d5fc --- /dev/null +++ b/src/network/netdev/vxlan.h @@ -0,0 +1,93 @@ +#pragma once + +/*** + This file is part of systemd. + + Copyright 2014 Tom Gundersen <teg@jklm.no> + + systemd is free software; you can redistribute it and/or modify it + under the terms of the GNU Lesser General Public License as published by + the Free Software Foundation; either version 2.1 of the License, or + (at your option) any later version. + + systemd is distributed in the hope that it will be useful, but + WITHOUT ANY WARRANTY; without even the implied warranty of + MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU + Lesser General Public License for more details. + + You should have received a copy of the GNU Lesser General Public License + along with systemd; If not, see <http://www.gnu.org/licenses/>. +***/ + +typedef struct VxLan VxLan; + +#include "in-addr-util.h" +#include "netdev/netdev.h" + +#define VXLAN_VID_MAX (1u << 24) - 1 + +struct VxLan { + NetDev meta; + + uint64_t id; + + int family; + union in_addr_union group; + + unsigned tos; + unsigned ttl; + unsigned max_fdb; + + uint16_t dest_port; + + usec_t fdb_ageing; + + bool learning; + bool arp_proxy; + bool route_short_circuit; + bool l2miss; + bool l3miss; + bool udpcsum; + bool udp6zerocsumtx; + bool udp6zerocsumrx; + bool remote_csum_tx; + bool remote_csum_rx; + bool group_policy; + + struct ifla_vxlan_port_range port_range; +}; + +DEFINE_NETDEV_CAST(VXLAN, VxLan); +extern const NetDevVTable vxlan_vtable; + +int config_parse_vxlan_group_address(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata); +int config_parse_port_range(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata); + +int config_parse_destination_port(const char *unit, + const char *filename, + unsigned line, + const char *section, + unsigned section_line, + const char *lvalue, + int ltype, + const char *rvalue, + void *data, + void *userdata); |