Repository navigation
feat: 802.1Q VLAN tag insert/strip support - #36
Merged
Merged
Conversation
…test Add full 802.1Q VLAN support to the socket layer: - VlanConfig type with VLAN ID (0-4094), PCP priority (0-7), DEI - build_udp_frame_into_vlan() for VLAN-tagged frame construction - parse_udp_packet/parse_udp_packet_ref transparently handle VLAN tags - verify_ipv4_checksum/verify_udp_checksum work on VLAN-tagged frames - process_frame_zerocopy dispatches ARP/ICMP/UDP correctly for tagged frames - ARP and ICMP parsers handle VLAN-tagged frames via detect_vlan() helper - Per-socket VLAN config via set_vlan()/vlan() and NetworkConfig::with_vlan() - UdpSocketBuilder applies VLAN config from NetworkConfig - 16 new unit tests covering build/parse/checksum/roundtrip/config Also adds Tier 4 jumbo frame echo integration test that exercises the jumbo frame feature (PR #32) with 1400/4000/8000-byte payloads on EC2. https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
[CI] Stage: DeployInfrastructure ready.
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
✅ Integration Tests Passed (Run 24312137815)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
…ring Add VlanMode enum matching Linux 8021q subinterface semantics: - Access: RX accepts untagged + matching VID (strip), TX sends untagged - Trunk: RX accepts allowed VID set + optional native VLAN, TX tags - PortTagging (default): RX only matching VID (strip), TX always tags RX filtering in process_frame_zerocopy drops frames before protocol dispatch (ARP/ICMP/UDP) based on the configured mode. TX path in send_to_addr respects mode: Access sends untagged, Trunk/PortTagging send tagged. 28 new unit tests covering all mode/frame combinations. https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
Measures process_frame_zerocopy throughput across all VLAN modes: - No config baseline: ~839K PPS (untagged) - PortTagging/Access/Trunk: ~760-853K PPS (<10% overhead) - DROP paths: >11M PPS (early reject) Run with: cargo test -p dpdk-stdlib-udp -- --nocapture vlan_pps_benchmark https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
[Perf] Stage: DeployDeploying |
Synthetic benchmark shows ~9-10% CPU overhead for VLAN-tagged frames. Hardware TRex results pending from perf-tests workflow. https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
Include VLAN PPS benchmark in every perf-test-log entry going forward. Runs in ~5s alongside cargo test, no hardware needed. https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
[CI] Stage: DeployInfrastructure ready.
|
[Perf] Stage: Instances Ready
|
[Perf] Stage: TRex ConfigStarting TRex configuration (MAC discovery + NIC binding)... |
[Perf] Stage: TRex Config OK
|
[Perf] Stage: TRex StartedTRex server running. Beginning benchmarks... |
[Perf] DUT ReadyDUT instance |
[Perf] Stage: Benchmark (1/4)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (2/4)Running |
[Perf] Benchmark Diag:
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
✅ Integration Tests Passed (Run 24312929132)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
Explains how DPDK port 0 is selected by default (PCI bus order), when explicit port selection is needed, and provides code examples for all backend configurations including DPDK port selection, AF_PACKET, MMAP, VLAN, and routing via the builder API. https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (3/4)Running |
[Perf] Benchmark Diag:
|
[CI] Stage: DeployInfrastructure ready.
|
[Perf] Benchmark Diag:
|
[Perf] Stage: Benchmark (4/4)Running |
[Perf] Benchmark Diag:
|
[Perf] Benchmark Diag:
|
[Perf] Diag: testpmd logtestpmd output (last 30 lines) |
No performance regression from 802.1Q VLAN changes. rust-dpdk delivers 690K RX at 700K PPS (1.3% drop) vs kernel's 399K (42.9% drop) — 1.73x throughput advantage. Within 0.6% of native C DPDK (695K RX). https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD
[Perf] Stage: Results[19:02:04] INFO Generating markdown summary... Performance Test Results — unknownCommit: 64B packets
512B packets
1400B packets
8500B packets
NIC Drops Instrumentation Self-CheckCompares
plain-rust Kernel NIC Drops (ethtool -S delta)ethtool snapshots not available — baseline or final file missing in |
[CI] Stage: SummarySome tests FAILED (exit code: 1). ARP seeding: kernel /proc/net/arp (automatic)
|
Integration Test Failure (Run 24313040593)Branch: No failure-summary.json found receiver-user-data.log (8237 bytes, last 80 lines)sender-user-data.log (8237 bytes, last 80 lines)receiver-console-output.log (32733 bytes, last 80 lines)sender-console-output.log (33280 bytes, last 80 lines)All instance-logs files |
❌ Integration Tests Failed (Run 24313040593)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
[CI] Stage: DeployInfrastructure ready.
|
[CI] Stage: SummaryAll tests PASSED. ARP seeding: kernel /proc/net/arp (automatic)
|
✅ Integration Tests Passed (Run 24314062570)Branch: Test Results
Application Logs (last 20 lines)receiver-echo-server.log sender-echo-server.log sender-test-client.log receiver-test-client-iperf.log sender-test-client-iperf.log Full Application Logs (last 200 lines each)receiver-echo-server.logsender-echo-server.logsender-test-client.logreceiver-test-client-iperf.logsender-test-client-iperf.log
|
gspivey
pushed a commit
that referenced
this pull request
Aug 16, 2026
ROADMAP item #36: EC2 integration test CI jobs with non-blocking TCP. Changes: 1. TCP tier orchestration (scripts/run-integration-tests.sh): - Added run_tcp_tier1, run_tcp_tier2, run_tcp_tier3 functions - TCP tier 1: handshake, echo, shutdown (DPDK<->DPDK) - TCP tier 2: retransmission under loss, flow control/persist probes - TCP tier 3: kernel interop (ncat/iperf3), std-parity - Extended --tier filter to accept tcp1, tcp1a, tcp2, tcp3 - TCP tiers run after UDP tier 4 when no filter is specified - tcp_cleanup_processes() kills stale DPDK processes before/between TCP tests to prevent /dev/vfio/noiommu-0 busy errors - DELETE_FAILED stack state recovery before CDK deploy 2. Non-blocking exit code (print_summary): - TCP test results (*tcp*.xml) excluded from exit code - TCP failures reported but don't block CI 3. Dedicated TCP workflow (scripts/ci/tcp-integration-tests.yml): - Ready-to-activate workflow with continue-on-error: true - Blocked by PAT lacking 'workflow' scope; staged for manual move 4. CI workflow fixes (.github/workflows/integration-tests.yml): - Exclude TCP tier XMLs from dorny fail-on-error reporter - Only UDP tier patterns in path glob (TCP handled separately) 5. tier1-tcp-echo.sh: accept server/client role aliases (was listener/sender), fixing 'Invalid role' errors that prevented TCP echo tests from running 6. ROADMAP housekeeping: ticked #35 (PR #117), tasks 15.2-15.4 Spec: .kiro/specs/tcp-support/ tasks 15.4, 16.1
gspivey
pushed a commit
that referenced
this pull request
Aug 16, 2026
ROADMAP item #36: EC2 integration test CI jobs with non-blocking TCP. Changes: 1. TCP tier orchestration (scripts/run-integration-tests.sh): - Added run_tcp_tier1, run_tcp_tier2, run_tcp_tier3 functions - TCP tier 1: handshake, echo, shutdown (DPDK<->DPDK) - TCP tier 2: retransmission under loss, flow control/persist probes - TCP tier 3: kernel interop (ncat/iperf3), std-parity - Extended --tier filter to accept tcp1, tcp1a, tcp2, tcp3 - TCP tiers run after UDP tier 4 when no filter is specified - tcp_cleanup_processes() kills stale DPDK processes before/between TCP tests to prevent /dev/vfio/noiommu-0 busy errors - DELETE_FAILED stack state recovery before CDK deploy 2. Non-blocking exit code (print_summary): - TCP test results (*tcp*.xml) excluded from exit code - TCP failures reported but don't block CI 3. Dedicated TCP workflow (scripts/ci/tcp-integration-tests.yml): - Ready-to-activate workflow with continue-on-error: true - Blocked by PAT lacking 'workflow' scope; staged for manual move 4. CI workflow fixes (.github/workflows/integration-tests.yml): - Exclude TCP tier XMLs from dorny fail-on-error reporter - Only UDP tier patterns in path glob (TCP handled separately) 5. tier1-tcp-echo.sh: accept server/client role aliases (was listener/sender), fixing 'Invalid role' errors that prevented TCP echo tests from running 6. ROADMAP housekeeping: ticked #35 (PR #117), tasks 15.2-15.4 Spec: .kiro/specs/tcp-support/ tasks 15.4, 16.1
gspivey
pushed a commit
that referenced
this pull request
Aug 16, 2026
ROADMAP item #36: EC2 integration test CI jobs with non-blocking TCP. Changes: 1. TCP tier orchestration (scripts/run-integration-tests.sh): - Added run_tcp_tier1, run_tcp_tier2, run_tcp_tier3 functions - TCP tier 1: handshake, echo, shutdown (DPDK<->DPDK) - TCP tier 2: retransmission under loss, flow control/persist probes - TCP tier 3: kernel interop (ncat/iperf3), std-parity - Extended --tier filter to accept tcp1, tcp1a, tcp2, tcp3 - TCP tiers run after UDP tier 4 when no filter is specified - tcp_cleanup_processes() kills stale DPDK processes before/between TCP tests to prevent /dev/vfio/noiommu-0 busy errors - DELETE_FAILED stack state recovery before CDK deploy 2. Non-blocking exit code (print_summary): - TCP test results (*tcp*.xml) excluded from exit code - TCP failures reported but don't block CI 3. Dedicated TCP workflow (scripts/ci/tcp-integration-tests.yml): - Ready-to-activate workflow with continue-on-error: true - Blocked by PAT lacking 'workflow' scope; staged for manual move 4. CI workflow fixes (.github/workflows/integration-tests.yml): - Exclude TCP tier XMLs from dorny fail-on-error reporter - Only UDP tier patterns in path glob (TCP handled separately) 5. tier1-tcp-echo.sh: accept server/client role aliases (was listener/sender), fixing 'Invalid role' errors that prevented TCP echo tests from running 6. ROADMAP housekeeping: ticked #35 (PR #117), tasks 15.2-15.4 Spec: .kiro/specs/tcp-support/ tasks 15.4, 16.1
gspivey
added a commit
that referenced
this pull request
Aug 18, 2026
## Summary Implements ROADMAP item #36: **dpdk-stdlib-tcp: EC2 CI jobs and remove continue-on-error gate**. Spec: `.kiro/specs/tcp-support/` tasks 15.4, 16.1 ### Changes **`integration-tests.yml`:** - Added `tcp-synthetic-perf` job: runs `tcp-synthetic-bench` (mock PacketBackend, no AWS needed) on every PR. Posts markdown results as a PR comment. Uploads results as artifacts with 30-day retention. - Added `tcp-integration-tests` job with `continue-on-error: true`: runs the full TCP test suite (tier1-tcp-handshake, tier1-tcp-echo, tier1-tcp-shutdown, tier2-tcp-retransmit, tier2-tcp-flow-control, tier3-tcp-kernel-interop, tier3-tcp-std-parity) on EC2. Uses `dorny/test-reporter` for PR checks UI. Posts structured PR comments with pass/fail/skip counts, app logs, and network state. Uploads JUnit XML + instance logs with 30-day retention. **`scripts/run-integration-tests.sh`:** - Added `--tier tcp-full` option for running the complete TCP test suite. - Added `run_tcp_full()` orchestrator function that sequences all TCP test tiers with proper ENI bind/unbind between each. - Added `_run_tcp_dpdk_pair()` helper (server/client roles) and `_run_tcp_dpdk_pair_lr()` helper (listener/sender roles) for DPDK↔DPDK tests. - Added `_run_tcp_kernel_interop()` helper for kernel↔DPDK tests. ### Task 16.1 (gate removal) The `continue-on-error: true` gate stays until ≥10/10 recent scheduled CI runs pass all MVP requirements. Recent runs show TCP tests are not yet passing consistently — gate removal will be a follow-up once the TCP stack matures. ### Tests - `cargo build && cargo test` pass locally (all 394+ tests, 0 failures) - This is a CI/scripts-only change — no Rust code modified --------- Co-authored-by: Agent Router <agent@agent-router.dev>
gspivey
added a commit
that referenced
this pull request
Aug 18, 2026
## Summary Marks ROADMAP item #36 (**dpdk-stdlib-tcp: EC2 CI jobs and remove continue-on-error gate**) as complete. ### Context The implementation was delivered in PR #120 (merged to development), which added: - `tcp-integration-tests` job with `continue-on-error: true` - PR comments with pass/fail/skip counts, app logs, network state - JUnit XML artifacts with 30-day retention - `dorny/test-reporter` for PR checks UI However, PR #120 did not update ROADMAP.md or tasks.md. This PR closes that gap. ### Changes - **ROADMAP.md**: Mark item #36 as `[x] Complete · PR: #120` - **tasks.md**: Mark task 15.4 as complete ### Task 16.1 (gate removal) Task 16.1 (remove `continue-on-error: true`) remains unchecked. The prerequisite (≥10/10 recent TCP CI runs passing) is not yet met — recent runs show TCP integration tests still failing. Gate removal will happen in a future session once the TCP stack matures. ### Tests - `cargo build && cargo test` pass (all tests ok, 0 failures) - Docs-only change — no Rust code modified Co-authored-by: Agent Router <agent@agentrouter.dev>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
set_vlan()or throughNetworkConfig::with_vlan().Changes
VLAN (802.1Q) Implementation
VlanConfigtype with builder pattern (VlanConfig::new(100).with_priority(3).with_dei(true))build_udp_frame_into_vlan()— frame builder with 4-byte VLAN tag insertiondetect_vlan()helper — determines L3 offset for both tagged and untagged framesparse_udp_packet/parse_udp_packet_refto handle VLAN-tagged frames (returnsvlan_id: Option<u16>)verify_ipv4_checksum/verify_udp_checksumfor correct VLAN offsetsprocess_frame_zerocopyto dispatch ARP/ICMP/UDP correctly for tagged framesparse_arp_packet) and ICMP parsers (parse_icmp_packet,parse_icmp_error) for VLANUdpSocket::set_vlan()/vlan()for per-socket VLAN configNetworkConfig::with_vlan()for builder-based configIntegration Test
tier4-jumbo-echo.sh— sends 1400B, 4000B, 8000B payloads via DPDK echorun-integration-tests.shas Tier 4Test plan
cargo buildpasses (all crates)cargo testpasses (321 tests, 0 failures, including 16 new VLAN tests)https://claude.ai/code/session_01Tumf1bXMixEaMKzgLvcBbD