[EVPN-MH] Integrate EVPN-MH with existing modules and add tests

[tahmed-dev]

Why I did it

This is part 2/2 of splitting the original EVPN-MH PR (#4262) into reviewable pieces.
Part 1 (#4608) landed the standalone EVPN-MH code (new orchs and headers added without touching existing files).
This PR integrates EVPN-MH with the existing modules (NeighOrch, FdbOrch, VxlanOrch, RouteOrch, MuxOrch, p4orch, fpmsyncd, etc.) and adds the mock/VS tests that exercise both the standalone and integration paths.

The test-infra prerequisites required to land these tests were merged separately as #4599.

How I did it

• Commit 1 — Integrate EVPN-MH with existing modules: wires the standalone EVPN-MH orchs (added in [EVPN-MH] Add standalone EVPN-MH code and tests #4608) into the existing data plane (FDB / neighbor / next-hop / VXLAN / mux / p4orch / fpmsyncd).
• Commit 2 — Add EVPN-MH tests and stabilization updates: adds new mock_tests (neighorch, vxlanorch, p4orch fake/mock helpers) and VS tests (test_sag.py, EVPN FDB/L3 VXLAN updates) and a few small stabilization tweaks needed for the new tests.

How to verify it

• make the swss debs and run mock_tests; test_sag.py and the new EVPN-MH cases under tests/mock_tests/{neighorch,vxlanorch}_ut.cpp should pass.
• Full VS suite locally and in CI.

Which release branch to backport (provide reason below if selected)

• 202012
• 202205
• 202211
• 202305
• 202311
• 202405
• 202411
• 202505
• 202511

Description for the changelog

[EVPN-MH] Integrate EVPN-MH with existing modules and add unit/VS tests.

Link to config_db schema for YANG module changes

N/A (YANG changes are tracked separately in sonic-buildimage PR #27543.)

A picture of a cute animal (not mandatory but encouraged)

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[Copilot]

Pull request overview

Note

Copilot was unable to run its full agentic suite in this review.

This PR expands EVPN/VXLAN and MUX test robustness, introduces Static Anycast Gateway (SAG) validation, and wires additional EVPN MH (DF/SHL/backup NHG) plumbing across orchagent, cfgmgr, and sync daemons.

Changes:

• Added SAG functional tests and SAG handling across IntfMgr/IntfsOrch, including link-local route programming tied to SAG MAC.
• Improved test stability by replacing fixed sleeps with state-based waits and adding ASIC state drain fixtures.
• Added EVPN MH / split-horizon / backup NHG support across fpmsyncd, portsorch, route/VRF orchestration, and mock/unit-test infrastructure.

Reviewed changes

Copilot reviewed 73 out of 73 changed files in this pull request and generated 3 comments.

Show a summary per file

File	Description
tests/test_sag.py	New functional tests covering SAG MAC set/unset, VLAN enable/disable, VRF behavior, and ASIC/kernel verification.
tests/test_mux_prefixroute.py	Reduces flakiness via wait-based FDB sync, neighbor drain waits, and ASIC cleanup fixture.
tests/test_inband_intf_mgmt_vrf.py	Simplifies VRF table empty wait helper to a direct implementation.
tests/test_evpn_l3_vxlan_p2mp.py	Refreshes baseline entries prior to VLAN interface creation to avoid cross-test pollution.
tests/test_evpn_l3_vxlan.py	Same baseline refresh prior to VLAN interface creation for determinism.
tests/test_evpn_fdb_p2mp.py	Adjusts synthetic FDB notification event type to MOVE.
tests/test_evpn_fdb.py	Adjusts synthetic FDB notification event type to MOVE.
tests/mock_tests/vxlanorch_ut.cpp	Expands VXLAN orch UT scaffolding and adds multiple new VXLAN unit tests.
tests/mock_tests/routeorch_ut.cpp	Adds EvpnMhOrch + IntfsOrch table vector wiring for mock route tests.
tests/mock_tests/qosorch_ut.cpp	Updates IntfsOrch construction for multi-table input (INTF + SAG).
tests/mock_tests/portsorch_ut.cpp	Adds EvpnMhOrch construction and updates IntfsOrch table wiring.
tests/mock_tests/neighorch_ut.cpp	Adds unit tests validating NeighOrch FDB add/delete/resolve behaviors.
tests/mock_tests/mux_rollback_ut.cpp	Adds missing gmock matcher import used by tests.
tests/mock_tests/mock_sai_tunnel.h	Converts mock SAI tunnel wrappers to extern declarations (implementation moved to .cpp).
tests/mock_tests/mock_orchagent_main.h	Adds EvpnMhOrch and L2NhgOrch globals for mock orchagent.
tests/mock_tests/mock_orchagent_main.cpp	Adds global SAG MAC variable initialization for mock orchagent.
tests/mock_tests/mock_orch_test.cpp	Ensures EvpnMhOrch is created early; updates IntfsOrch ctor signature usage.
tests/mock_tests/mock_hiredis.cpp	Implements redisFree() to close fd/free allocations in mock hiredis context.
tests/mock_tests/intfsorch_ut.cpp	Updates IntfsOrch construction to accept table vector (INTF + SAG).
tests/mock_tests/fpmsyncd/ut_helpers_fpmsyncd.cpp	Extends mocked ifindex→name mapping for EVPN MH tests.
tests/mock_tests/flowcounterrouteorch_ut.cpp	Updates IntfsOrch construction to accept table vector (INTF + SAG).
tests/mock_tests/flexcounter_ut.cpp	Updates IntfsOrch construction to accept table vector (INTF + SAG).
tests/mock_tests/fdborch/flush_syncd_notif_ut.cpp	Adds missing orch dependencies; updates remote FDB data model expectations.
tests/mock_tests/check.h	Fixes signed/unsigned compare by casting act_len/exp_len to size_t.
tests/mock_tests/bufferorch_ut.cpp	Updates IntfsOrch construction to accept table vector (INTF + SAG).
tests/mock_tests/buffermgrdyn_ut.cpp	Ensures WarmStart flag is reset in TearDown to avoid cross-test impact.
tests/mock_tests/aclorch_ut.cpp	Updates IntfsOrch construction to accept table vector (INTF + SAG).
tests/mock_tests/Makefile.am	Adds new test binaries/sources (fdbsyncd UTs, EVPN MH, SHL, mock SAI files, helpers).
run-gtest-suite.py	Changes RLIMIT_NOFILE adjustment logic (now unconditionally imports resource).
orchagent/vxlanorch.h	Adds DIP cleanup APIs (cleanupDynamicDIPTunnel/eraseRemoteEndPoint).
orchagent/vxlanorch.cpp	Adjusts DIP endpoint lifecycle, improves logging, and fixes VNI L2/L3 mapping ordering.
orchagent/vrforch.h	Adds VRF ID enumeration helper and markVniAsL3 helper.
orchagent/vrforch.cpp	Moves link-local route management to RouteOrch via Directory lookup; adds per-VRF EUI64 /128 CPU route.
orchagent/routeorch.h	Changes getLinkLocalEui64Addr() to accept a MAC parameter.
orchagent/routeorch.cpp	Updates link-local route creation to use the passed MAC rather than global-only.
orchagent/request_parser.cpp	Broadens colon-joined key parsing to support REQ_T_STRING final key item types.
orchagent/portsorch.h	Adds VLAN member lookup and L2 NHG add/remove APIs.
orchagent/portsorch.cpp	Adds VLAN member lookup, L2 NHG bridge-port creation, EVPN DF attributes, and tunnel oper status updates.
orchagent/port.h	Adds Port::NEXTHOP_GROUP type and storage for next-hop group OID.
orchagent/p4orch/tests/test_main.cpp	Switches to mock RouteOrch, wires global gMacAddress and creates a RouteOrch instance for tests.
orchagent/p4orch/tests/mock_sai_tunnel.h	Adds extern mock_sai_tunnel pointer and formatting fixes.
orchagent/p4orch/tests/mock_sai_tunnel.cpp	Formatting + style fixes; ensures mock_sai_tunnel global is defined.
orchagent/p4orch/tests/fake_routeorch.cpp	Replaces stub RouteOrch with a more complete mock implementation for newer interfaces.
orchagent/p4orch/tests/Makefile.am	Reorders mock_sai_tunnel.cpp in sources list (no functional change).
orchagent/orchdaemon.h	Adds EvpnMhOrch/L2NhgOrch/ShlOrch includes.
orchagent/orchdaemon.cpp	Creates EvpnMhOrch earlier, introduces L2NhgOrch/ShlOrch, updates orch init ordering.
orchagent/neighorch.h	Adds deletion_pending flag and FDB event handlers + VRF-aware getNeighborEntry overload.
orchagent/neighorch.cpp	Implements FDB add/delete/resolve handlers; adjusts neighbor programming logic and deletion notify ordering.
orchagent/muxorch.cpp	Makes route creation accept const prefix; adds item-not-found create fallback and improves prefix-route deletion logic.
orchagent/intfsorch.h	Changes ctor to accept multi-table list; adds SAG state fields and SAG route helpers.
orchagent/intfsorch.cpp	Adds SAG table handling, per-interface MAC tracking, and link-local route programming tied to SAG enablement.
orchagent/fdborch.h	Extends remote FDB destination model + adds flushAllFDBEntries API and per-port cache structures.
orchagent/Makefile.am	Adds new orchagent sources: evpnmhorch.cpp, l2nhgorch.cpp, shlorch.cpp.
neighsyncd/neighsyncd.cpp	Updates NeighSync ctor to include APP_DB connector.
neighsyncd/neighsync.h	Adds route table + netlink cache/socket members and interface-name helper.
neighsyncd/neighsync.cpp	Adds netlink cache/socket lifecycle, VRF master lookup, and host route deletion before neighbor add.
fpmsyncd/routesync.h	Adds EVPN MH PS tables + raw handlers (SHL/DF/backup NHG/TC filter) and tuple wrapper includeEmptyFields.
fpmsyncd/routesync.cpp	Adds EVPN MH parsing (via/tc filter/FPM private types), adjusts VXLAN encap parsing, adds RouteSync destructor.
fpmsyncd/fpmsyncd.cpp	Registers RTM_NEWTFILTER/RTM_DELTFILTER with NetDispatcher.
fpmsyncd/fpmlink.cpp	Treats TFILTER + RTM_FPM private types as raw processing; null-guards RouteSync callbacks.
fpmsyncd/fpm/fpm.h	Defines FPM private RTM_FPM_* message types and EVPN MH message structs/attributes.
fdbsyncd/fdbsyncd.cpp	Moves neigh/nexthop handling to raw netlink handlers; adds RTNLGRP_NEXTHOP dump/subscribe.
fdbsyncd/fdbsync.h	Extends FdbSync interface for raw messages, L2 NHG cache, and richer destination typing.
cfgmgr/vxlanmgr.h	Splits ProducerStateTable vs Table for APP_VXLAN_TUNNEL updates.
cfgmgr/vxlanmgr.cpp	Uses ProducerStateTable for tunnel updates; adds udp6zerocsumrx behavior for IPv6 VTEPs; delays NVO until tunnel exists.
cfgmgr/teammgr.h	Adds system MAC setter and kernel update helper for LAG.
cfgmgr/teammgr.cpp	Implements kernel MAC update via libnl and propagates system_mac to APP/STATE DB.
cfgmgr/intfmgrd.cpp	Adds CFG_SAG_TABLE_NAME subscription and initializes global MAC variables from DEVICE_METADATA.
cfgmgr/intfmgr.h	Adds SAG producers/tables and helpers for SAG FDB + interface state changes.
cfgmgr/intfmgr.cpp	Implements SAG global MAC + VLAN per-interface SAG MAC switching, including FDB programming and APP_DB SAG updates.

Comments suppressed due to low confidence (4)

run-gtest-suite.py:1

• import resource is now unconditional at module import time, which will raise ImportError on platforms without the POSIX resource module (the comment explicitly mentions Windows). To keep the script portable, remove the unconditional import and keep the guarded try/except ImportError import (or import inside the try block only), then gate the RLIMIT_NOFILE logic on whether the import succeeded.
  orchagent/portsorch.cpp:1
• This change introduces a call to FdbOrch::flushAllFDBEntries(...). In this PR diff set, only the declaration is added (in orchagent/fdborch.h), but there is no corresponding implementation shown (e.g., in orchagent/fdborch.cpp). If the implementation is missing, this will fail to link. Add/verify the concrete implementation and ensure it correctly flushes both dynamic and static entries for the given bridge port.
  orchagent/request_parser.cpp:1
• The logic that follows this condition is described (in the nearby comment) as assembling removed items into an IPv6 address. After extending the condition to REQ_T_STRING, that comment is no longer accurate. Update the comment to reflect that the code is now also used to re-join arbitrary string key components that may include ':' (not just IPv6 parsing).
  orchagent/vxlanorch.h:1
• The newly added APIs (and a couple of existing ones) take std::string by value, which causes unnecessary copies on hot paths and is inconsistent with other APIs in this area. Prefer const std::string& remote_vtep (and similarly for dip) unless ownership transfer is intended.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[tahmed-dev]

@prsunny @pbrisset — flagging reviewer focus areas where this PR alters existing code paths (vs. purely additive code). Per-file/line comments below.

[tahmed-dev]

Init order changed: EvpnMhOrch constructed before FdbOrch and inserted between gPortsOrch and gBufferOrch in m_orchList; L2NhgOrch appended at end; gShlOrch appended after VXLAN orchs; IntfsOrch ctor signature changed to add APP_SAG_TABLE_NAME. Please verify warm-boot replay order and that ES/DF state is available before FdbOrch/PortsOrch consume it.

[tahmed-dev]

Code is provided to subdue this concern by constructing EVPN-MH before FDB processing and placing it immediately after PortsOrch in the warm-restart list, so ES/DF state is available before FdbOrch consumes FDB work. L2 NHG remains in the orch list and the SAG table is added at the same IntfsOrch priority as INTF_TABLE.

// Create EvpnMhOrch early so its ES/DF state is available when PortsOrch
// processes bridge ports and VLAN members (fixes warm boot ordering)
gEvpnMhOrch = new EvpnMhOrch(evpn_df_es_table_connectors);

m_orchList = { gSwitchOrch, gCrmOrch, gPortsOrch, gEvpnMhOrch,
               gBufferOrch, gFlowCounterRouteOrch, gIntfsOrch,
               gNeighOrch, /* ... */, gStpOrch, gL2NhgOrch };

vector<table_name_with_pri_t> intf_tables = {
    { APP_INTF_TABLE_NAME,  IntfsOrch::intfsorch_pri},
    { APP_SAG_TABLE_NAME,   IntfsOrch::intfsorch_pri}
};

[tahmed-dev]

FDB state machine changes: storeFdbEntryState() on MAC-move now explicitly m_entries.erase(entry) before re-insert; remote_ip replaced by dest_type/dest_value (IFNAME). clearFdbEntry() signature changed to (entry, fdbData); per-port FDB counter decrement is now conditional on getPortByBridgePortId() (previously unconditional). handleSyncdFlushNotif() has three hunks changing bv_id/bridge_port_id resolution during flush. Please double-check counter-decrement parity in every pre-existing path and that flush behavior is unchanged for non-EVPN FDB.

[tahmed-dev]

Code is provided to keep the legacy FDB paths idempotent while allowing the EVPN-MH destination metadata. The branch trace confirms the existing non-EVPN flush path still requires the same bv_id and bridge_port_id match before clearing, and the removal path keeps counter updates tied to resolved bridge ports instead of touching unknown ports.

if (curr->first.bv_id == bv_id && curr->second.bridge_port_id == bridge_port_id)
{
    if (curr->second.sai_fdb_type == sai_fdb_type &&
        (curr->first.mac == mac || mac == flush_mac) && curr->second.is_flush_pending)
    {
        clearFdbEntry(curr->first, curr->second);
    }
}

if (bridge_port_id &&
    !m_portsOrch->getPortByBridgePortId(bridge_port_id, update.port))
{
    if (type == SAI_FDB_EVENT_FLUSHED)
    {
        return;
    }
}

[tahmed-dev]

Neighbor add/remove altered for ES/DF: addNeighbor() integrates ES/DF check (3 hunks) and may now refuse to program a neighbor that previously would have been programmed; removeNeighbor() gains a disable bool parameter (all call sites updated); both getNeighborEntry(NextHopKey) and getNeighborEntry(IpAddress) modified. Please check legacy callers' expectations when a neighbor is 'disabled but not removed' and that isHwConfigured() is cleared correctly.

[tahmed-dev]

Code is provided to keep the disabled-neighbor state explicit instead of deleting the cache entry. The disable path returns early for neighbors that are already absent from hardware, and removeNeighbor(..., true) clears hw_configured while preserving the cache entry for later re-enable.

if (!isHwConfigured(neighborEntry))
{
    SWSS_LOG_INFO("Neighbor %s is not programmed to HW", neighborEntry.ip_address.to_string().c_str());
    return true;
}

return removeNeighbor(ctx, true);

/* Do not delete entry from cache if its disable request */
if (disable)
{
    m_syncdNeighbors[neighborEntry].hw_configured = false;
    return true;
}

[tahmed-dev]

MUX neighbor update touched by EVPN-MH: set_route() adds an extra branch; MuxNbrHandler::update() and MuxPrefixBasedNbrHandler::update() now factor EVPN-MH into SAI next-hop / tunnel_id during active/standby transitions. Please confirm standby→active and prefix-route fast-path are bit-identical when no ES is configured.

[tahmed-dev]

Code is provided to keep the non-EVPN MUX fast path explicit. The set-route fallback only creates a route when the prefix-route path sees SAI_STATUS_ITEM_NOT_FOUND; otherwise the existing set failure handling is unchanged. Delete handling removes only tracked prefix routes, or preserves the prior standby cleanup path.

sai_status_t status = sai_route_api->set_route_entry_attribute(&route_entry, &route_attr);
if (status == SAI_STATUS_ITEM_NOT_FOUND && mux_prefix_route)
{
    return create_route(pfx, next_hop_id);
}

if (neighbors_.find(nh.ip_address) != neighbors_.end() || state == MuxState::MUX_STATE_STANDBY)
{
    remove_route(pfx);
}

if (state == MuxState::MUX_STATE_STANDBY)
{
    updateTunnelRoute(nh, false);
}

[tahmed-dev]

Netlink decode rewritten: getEvpnNextHop() rewritten; new rtattr parsers added; onMsg() decode path gains branches for L3 VNI / EVPN-MH attributes. Please verify non-EVPN route decoding is bit-for-bit identical to previous master.

[tahmed-dev]

Code is provided to keep legacy route decoding separate from the EVPN/LWT encap additions. The existing RTA_GATEWAY path still determines IPv4/IPv6 from payload size; the new RTA_VIA branch is only used when the gateway attribute is absent. Encapsulation parsing returns false if VNI, interface, or RMAC is missing.

if (gate)
{
    if (RTA_PAYLOAD(tb[RTA_GATEWAY]) <= IPV4_MAX_BYTE)
    {
        memcpy(gateaddr, gate, IPV4_MAX_BYTE);
        gw_af = AF_INET;
    }
    else
    {
        memcpy(ipv6_address.s6_addr, gate, IPV6_MAX_BYTE);
        gw_af = AF_INET6;
    }
}
else if (tb[RTA_VIA])
{
    via = (struct rtvia *)RTA_DATA(tb[RTA_VIA]);
    /* parse VIA only when RTA_GATEWAY is not present */
}

if (encap_value == 0 || !(vlan.compare(ifname_unknown)) || MacAddress(rmac) == MacAddress("00:00:00:00:00:00"))
{
    return false;
}

[tahmed-dev]

Interface bring-up changes: setIntfIp() adds a SAG path, setIntfMpls() adds an early-return branch, doIntfGeneralTask() has 4 new hunks. Please check idempotence on re-apply.

[tahmed-dev]

Code is provided to keep interface re-apply idempotent. SAG logic only runs when the static_anycast_gateway field is present; otherwise the legacy default mac_addr tuple is still pushed. SAG enable/disable reuses replace/delete FDB operations so repeated application lands in the same state.

if (!sag.empty())
{
    if (!alias.compare(0, strlen(VLAN_PREFIX), VLAN_PREFIX))
    {
        if (sag == "true")
        {
            m_sagIntfList[alias] = true;
            setSagFdbEntry("replace", alias, gwmac);
            data.push_back(FieldValueTuple("mac_addr", gwmac));
        }
        else if (sag == "false")
        {
            m_sagIntfList[alias] = false;
            setSagFdbEntry("del", alias, gSagMacAddress.to_string());
            data.push_back(FieldValueTuple("mac_addr", MacAddress().to_string()));
        }
    }
}
else
{
    data.push_back(FieldValueTuple("mac_addr", MacAddress().to_string()));
}

[tahmed-dev]

LAG task + learn-mode: doLagTask() modifications and a new setLagLearnMode() block (~+105 lines). Please confirm legacy non-EVPN LAG bring-up sequence is unchanged.

[tahmed-dev]

Code is provided to keep legacy LAG bring-up sequencing intact. The existing create/admin/MTU/learn-mode/TPID flow still runs first, and the new system MAC update is conditional on system_mac being present. The kernel update uses a valid flags value for rtnl_link_change() and only writes APP_DB/STATE_DB after kernel success.

setLagAdminStatus(alias, admin_status);
setLagMtu(alias, mtu);
if (!learn_mode.empty())
{
    setLagLearnMode(alias, learn_mode);
}
if (!tpid.empty())
{
    setLagTpid(alias, tpid);
}
if (!sys_mac.empty())
{
    setLagSysmac(alias, sys_mac);
}

if (rtnl_link_change(sockk, orig_link, link, 0) < 0)
{
    return -1;
}

m_appLagTable.set(alias, fvs);
m_stateLagTable.set(alias, fvs);

[tahmed-dev]

VXLAN create / tunnel-create flow modified: please re-verify idempotency on warm boot.

[tahmed-dev]

Code is provided to keep VXLAN tunnel create and NVO replay idempotent. Tunnel create stores the same APP_DB payload through ProducerStateTable and refreshes the cache deterministically; NVO creation waits until the referenced tunnel is active and present in APP_DB; IPv6-only options are appended only when the source address is IPv6.

m_appVxlanTunnelTableProducer.set(vxlanTunnelName, kfvFieldsValues(t));
m_vxlanTunnelCache[vxlanTunnelName] = tuncache;

if (!m_appVxlanTunnelTable.get(value, fv))
{
    SWSS_LOG_WARN("NVO %s creation delayed. VTEP %s not found", EvpnNvoName.c_str(), value.c_str());
    return false;
}

if (inet_pton(AF_INET6, src_ip.c_str(), &addr6) == 1)
{
    link_add_cmd += " udp6zerocsumrx";
}

[tahmed-dev]

Small struct addition — safe, but this is a widely-included header; please confirm no downstream #include cycle or size-sensitive consumer is affected.

[tahmed-dev]

Code is provided with a minimal Port extension: one enum value and one object-id field, both in the existing Port structure and both defaulted through the existing zero-initialization style. This does not add a new include dependency or introduce a downstream include cycle.

enum Type
{
    PHY,
    MGMT,
    LOOPBACK,
    VLAN,
    LAG,
    TUNNEL,
    SUBPORT,
    SYSTEM,
    NEXTHOP_GROUP,
    UNKNOWN
};

sai_object_id_t m_nexthop_group_id = 0;

[tahmed-dev]

New sources added — build-only; please sanity-check the build rules / link order.

[tahmed-dev]

Code is provided in the build rules by adding the new sources to the existing orchagent source list before the dependent users are linked. The Trixie swss package build completed with these objects in the standard build path.

            routeorch.cpp \
            mplsrouteorch.cpp \
            neighorch.cpp \
            intfsorch.cpp \
            evpnmhorch.cpp \
            l2nhgorch.cpp \
            port/port_capabilities.cpp \

            high_frequency_telemetry/hftelgroup.cpp \
            shlorch.cpp

Validation: Trixie swss build completed with PASS: 770/0 fail, PASS: 863/0 fail, and PASS: 70/0 fail.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[tahmed-dev]

/azpw run

[mssonicbld]

⚠️ Notice: /azpw run only runs failed jobs now. If you want to trigger a whole pipline run, please rebase your branch or close and reopen the PR.
💡 Tip: You can also use /azpw retry to retry failed jobs directly.

Retrying failed(or canceled) jobs...

[mssonicbld]

Retrying failed(or canceled) stages in build 1124559:

✅Stage BuildTrixie:

• Job amd64: retried.
• Job arm64: retried.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[tahmed-dev]

/azpw run

[mssonicbld]

⚠️ Notice: /azpw run only runs failed jobs now. If you want to trigger a whole pipline run, please rebase your branch or close and reopen the PR.
💡 Tip: You can also use /azpw retry to retry failed jobs directly.

Retrying failed(or canceled) jobs...

[mssonicbld]

Retrying failed(or canceled) stages in build 1125152:

✅Stage Test:

• Job vstest: retried.

✅Stage BuildAsan:

• Job amd64: retried.

[prsunny]

Why change the signature here? As much as possible, could you limit the changes?

[tahmed-dev]

This parameterization is needed for SAG (Static Anycast Gateway): the SAG VLAN interface paths in intfsorch.cpp now call getLinkLocalEui64Addr(m_sagMac) so the link-local EUI64 is derived from the SAG MAC instead of the global switch MAC. The existing callers in routeorch.cpp and vrforch.cpp still pass gMacAddress, so the non-SAG path is unchanged — the only delta is adding the mac parameter.

[prsunny]

What is this change?

[tahmed-dev]

This was unintended churn — it changed the log message to use a local neighbor_mac instead of m_syncdNeighbors[neighborEntry].mac. I have reverted it, so this line now matches master.

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[mssonicbld]

/azp run

[azure-pipelines]

Azure Pipelines successfully started running 1 pipeline(s).

[tahmed-dev]

/azpw run

[mssonicbld]

⚠️ Notice: /azpw run only runs failed jobs now. If you want to trigger a whole pipline run, please rebase your branch or close and reopen the PR.
💡 Tip: You can also use /azpw retry to retry failed jobs directly.

Retrying failed(or canceled) jobs...