feat: S28.09 FreeRtosLwip LAN9118 netif + NO_SYS=0 UDP on QEMU#476
Merged
Conversation
Grow the S28.07 link-probe into a runnable FreeRTOS+lwIP BDD target on QEMU mps2-an385: a hand-written LAN9118 lwIP netif (netif/EthernetIf.c) over the vendored Arm smsc9220 driver (Apache-2.0), tcpip_init + static IP bring-up on the tcpip thread, and the SolidSyslog UDP pipeline wired through the tcpip_callback marshal (S28.06). Verified on QEMU: boot, ARP, and a correct RFC 5424 datagram to 10.0.2.2:5514, clean teardown, no PlusTcp symbols. WIP: oracle BDD run + CI lanes still to come. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Add the syslog-ng-freertos-lwip / behave-freertos-lwip compose pair (mirrors the +TCP pair, points at the lwIP ELF, UDP-only tag filter with a @freertoslwipwip escape hatch) and the advisory bdd-freertos-qemu-lwip CI job (not in summary.needs). Grow build-freertos-target-lwip to upload the BDD tunables mirror; refresh the link-probe wording to BDD target across ci.yml, the top-level CMake status, and the README. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
…rmat The LwipRaw datagram sends PBUF_REF packets pointing at a transient buffer, so a first send that hit an ARP miss had its queued copy reference freed memory and was dropped (confirmed on the wire: seqId=1 lost, seqId=2 delivered). Warm up the gateway ARP at bring-up (etharp_request) and block the interactive task — never the tcpip thread — until the cache populates before the pipeline goes live. Verified: both seqId=1 and seqId=2 now reach the wire. Also move the vendored Arm smsc9220 driver into netif/smsc9220/ behind a DisableFormat .clang-format so the analyze-format lane does not reflow third-party code, and clang-format the hand-written netif / main / config files. Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Contributor
|
No actionable comments were generated in the recent review. 🎉 ℹ️ Recent review info⚙️ Run configurationConfiguration used: defaults Review profile: CHILL Plan: Pro Run ID: 📒 Files selected for processing (4)
✅ Files skipped from review due to trivial changes (1)
🚧 Files skipped from review as they are similar to previous changes (3)
📝 WalkthroughWalkthroughThis PR converts the FreeRTOS+lwIP BDD target from a CI-only link‑probe into a runnable NO_SYS=0 QEMU target: adds a vendored LAN9118 driver, lwIP netif wrapper, expanded FreeRTOS/lwIP threading config, a SolidSyslog interactive BDD runtime with ARP warm‑up, CMake wiring, CI lane, and docs/DEVLOG entries. ChangesFreeRTOS+lwIP BDD Target
Sequence DiagramsequenceDiagram
participant Developer
participant CI
participant QEMU as FreeRTOS/QEMU target
participant lwIP as tcpip thread
participant Driver as SMSC9220 driver
participant Oracle as syslog-ng
Developer->>CI: push PR (build & artifacts)
CI->>QEMU: upload/download ELF & tunables
CI->>Oracle: start syslog-ng-freertos-lwip
CI->>QEMU: start behave-freertos-lwip (runs tests)
QEMU->>lwIP: tcpip_init, netif_add, ARP warm-up
lwIP->>Driver: linkoutput/send_by_chunks (TX)
Driver->>Oracle: network packets delivered to syslog-ng
Oracle->>CI: test results (JUnit)
Estimated code review effort🎯 4 (Complex) | ⏱️ ~75 minutes Possibly related issues
Possibly related PRs
Poem
🚥 Pre-merge checks | ✅ 4 | ❌ 1❌ Failed checks (1 warning)
✅ Passed checks (4 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
Contributor
☀️ Quality Summary 🚦 build-linux-gcc: 100% successful (✔️ 1344 passed) Created by Quality Monitor v1.14.0 (#f3859fd). More details are shown in the GitHub Checks Result. |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 12
🧹 Nitpick comments (1)
Bdd/Targets/FreeRtosLwip/netif/EthernetIf.c (1)
227-230: ⚡ Quick winPrefer
smsc9220_get_interrupt()over the localextern get_irq_status()inEthernetISR.
smsc9220_eth_drv.hexposesint smsc9220_get_interrupt(dev, source), which returnsGET_BIT(register_map->irq_status, source); since the ISR already usesSMSC9220_INTERRUPT_RX_STATUS_FIFO_LEVEL, replace theget_irq_status(dev) & rxFifoStatusBitcheck withsmsc9220_get_interrupt(dev, SMSC9220_INTERRUPT_RX_STATUS_FIFO_LEVEL) != 0and drop the localexternto avoid coupling to the internal symbol.🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@Bdd/Targets/FreeRtosLwip/netif/EthernetIf.c` around lines 227 - 230, In EthernetISR replace the use of the local extern get_irq_status(dev) check with the public API: call smsc9220_get_interrupt(dev, SMSC9220_INTERRUPT_RX_STATUS_FIFO_LEVEL) and test it != 0, and remove the local extern uint32_t get_irq_status(...) declaration; this keeps the ISR using the driver-exported smsc9220_get_interrupt function and avoids coupling to the internal get_irq_status symbol.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In @.github/workflows/ci.yml:
- Around line 1157-1158: The workflow uses the mutable reference
dorny/test-reporter@v3 (e.g., the step named "BDD Test Report (FreeRTOS+lwIP)")
which should be pinned to a specific commit SHA; update every occurrence of
dorny/test-reporter@v3 in the CI workflow to the full commit SHA (the same
pattern used for other actions) so the action reference is immutable and
reproducible.
- Line 1133: The checkout step in the bdd-freertos-qemu-lwip job is leaving Git
credentials inside the workspace that get bind-mounted into containers; update
the actions/checkout@... step (the checkout action invocation) to include
persist-credentials: false so credentials are not written into the checked-out
.git directory before the docker compose step mounts the repo.
In `@Bdd/Targets/FreeRtosLwip/main.c`:
- Around line 567-587: TeardownAll uses a fixed 20ms vTaskDelay to hope
ServiceTask sees solidSyslogTeardown before lifecycleMutex is destroyed; instead
ensure the service has actually exited (or been signalled to exit) before
destroying lifecycleMutex. Modify TeardownAll to set solidSyslogTeardown, then
notify the service task (e.g., xTaskNotify or give a semaphore) and block
waiting for a confirmation from the service (e.g., service clears
serviceTaskHandle or gives a "terminated" semaphore/task notification) or poll
serviceTaskHandle in a loop with vTaskDelay and a sensible timeout; only after
receiving the explicit confirmation that ServiceTask has finished should you
call SolidSyslogFreeRtosMutex_Destroy(lifecycleMutex). Use the existing symbols
solidSyslogTeardown, serviceTaskHandle, lifecycleMutex and TeardownAll to
implement this handshake.
- Around line 368-370: When updating the "host" via TryUpdateString(host,
sizeof(host), value) in main.c, also increment the endpointVersion so
GetEndpointVersion() changes and the new host takes effect immediately; modify
the branch handling strcmp(name, "host") to detect a successful update
(TryUpdateString returned true) and bump endpointVersion (the variable used by
GetEndpointVersion()) after the update, similar to how port changes trigger the
version bump.
- Around line 204-206: Check the return values from both xTaskCreate calls (the
InteractiveTask and ServiceTask creations) instead of ignoring them; if either
returns an error (not pdPASS) log/report the failure and avoid calling
vTaskStartScheduler (e.g., return or loop/error-handle), and ensure
serviceTaskHandle is only used if ServiceTask creation succeeded. Update the
xTaskCreate calls in main (InteractiveTask, ServiceTask,
SERVICE_TASK_STACK_DEPTH, INTERACTIVE_TASK_STACK_DEPTH, and serviceTaskHandle)
to capture and check their return values and handle a failed creation path
before vTaskStartScheduler is invoked.
- Around line 256-282: The code incorrectly uses tcpip_callback which only
queues work; change both WarmUpGatewayArp and LwipTcpipMarshal to use
tcpip_callback_wait (or an equivalent completion API) so the caller blocks until
the posted callback finishes; in WarmUpGatewayArp ensure the boolean resolved is
safe to pass (stack-to-callback race) by either allocating a stable completion
object or using tcpip_callback_wait so the callback writes to resolved before
the function reads it; replace tcpip_callback(callback, context) calls in
LwipTcpipMarshal and the WarmUpGatewayArp loop with tcpip_callback_wait to
guarantee synchronous behavior.
In `@Bdd/Targets/FreeRtosLwip/netif/EthernetIf.c`:
- Around line 197-219: EthernetIf_LowLevelInput currently returns NULL for
overlength frames and leaves the RX FIFO intact; change it so that when
smsc9220_peek_next_packet_size() returns length > sizeof(EthernetIf_RxBuffer)
you explicitly drain that frame by calling smsc9220_receive_by_chunks() (into a
temporary discard buffer or in chunks) and update stats (e.g.,
LINK_STATS_INC(link.drop) and LINK_STATS_INC(link.memerr) as appropriate) before
returning NULL; keep the normal path for valid lengths that allocate a pbuf, and
ensure this drain behavior mirrors what
empty_rx_fifo()/smsc9220_receive_by_chunks() expect so the RX FIFO level
interrupt can deassert.
In `@Bdd/Targets/FreeRtosLwip/netif/smsc9220/smsc9220_eth_drv.c`:
- Around line 319-336: The code is prepending a pad DWORD even when the chunk is
already 4-byte aligned; change the logic so when remainder_bytes == 0 you do not
build or write a leading padded word. Specifically, in the block that computes
remainder_bytes and filler_bytes and fills tx_data_port_tmp_ptr (using
remainder_bytes, filler_bytes, tx_data_port_tmp_ptr, tx_data_port_tmp,
register_map->tx_data_port, size_bytes and data), skip the padding loop and the
write to register_map->tx_data_port for aligned chunks — only write actual data
words directly when remainder_bytes == 0; when remainder_bytes != 0 keep the
existing filler loop but only subtract remainder_bytes from size_bytes and
advance data when you actually consumed those start bytes. Also apply the same
fix at the other location that uses current_size % 4 (the 1150-1155 block) so
aligned segments are not prepended with a dummy DWORD.
- Around line 455-468: The busy-bit poll in smsc9220_mac_regwrite is incorrectly
masking register_map->mac_csr_cmd with the boolean result of GET_BIT, causing
the loop to test bit 0 instead of the BUSY bit; update the loop condition to
directly test the BUSY bit (either by using GET_BIT(register_map->mac_csr_cmd,
MAC_CSR_CMD_BUSY_INDEX) as the boolean expression or by masking
register_map->mac_csr_cmd with (1u << MAC_CSR_CMD_BUSY_INDEX)) so the loop only
exits when the BUSY bit is cleared or time_out reaches zero; change the
condition in the do/while to use one of these correct checks and keep the rest
of the smsc9220_mac_regwrite logic intact.
- Around line 855-859: smsc9220_check_id currently returns 0/1 causing callers
(like smsc9220_init) to interpret failure as SMSC9220_ERROR_TIMEOUT; change its
return to the proper smsc9220_error_t values: check the chip ID using
GET_BIT_FIELD(id, CHIP_ID_MASK, CHIP_ID_POS) == CHIP_ID and return the success
enum (e.g., SMSC9220_ERROR_NONE or SMSC9220_OK) on match and a specific chip-id
error enum (e.g., SMSC9220_ERROR_CHIP_ID or SMSC9220_ERROR_ID_MISMATCH) on
mismatch so callers receive the correct error code. Ensure the function
signature remains int -> smsc9220_error_t (or update signature to
smsc9220_error_t) and update any callers if you change the return type.
In `@Bdd/Targets/FreeRtosLwip/README.md`:
- Around line 27-29: The README references incorrect vendored driver
paths—update the mention of `netif/smsc9220_eth_drv.{c,h}` and
`netif/smsc9220_emac_config.h` to reflect the actual directory layout under
`netif/smsc9220/` (for example, `netif/smsc9220/smsc9220_eth_drv.c`,
`netif/smsc9220/smsc9220_eth_drv.h`, and
`netif/smsc9220/smsc9220_emac_config.h`) so the documented paths match the
repository structure in README.md.
In `@DEVLOG.md`:
- Around line 12986-12993: Update the DEVLOG wording to avoid implying an lwIP
API called tcpip_callback_with_block: explain that the adapters use the blocking
marshal wrapper implemented by LwipTcpipMarshal (installed via
SolidSyslogLwipRaw_SetMarshal), and that LwipTcpipMarshal internally calls the
lwIP API tcpip_callback(...) to achieve the synchronous/blocking behavior;
either rename the informal “tcpip_callback_with_block” to “LwipTcpipMarshal
(tcpip_callback-based blocking marshal)” or add an explicit mapping sentence
linking the informal term to the LwipTcpipMarshal implementation and its use of
tcpip_callback.
---
Nitpick comments:
In `@Bdd/Targets/FreeRtosLwip/netif/EthernetIf.c`:
- Around line 227-230: In EthernetISR replace the use of the local extern
get_irq_status(dev) check with the public API: call smsc9220_get_interrupt(dev,
SMSC9220_INTERRUPT_RX_STATUS_FIFO_LEVEL) and test it != 0, and remove the local
extern uint32_t get_irq_status(...) declaration; this keeps the ISR using the
driver-exported smsc9220_get_interrupt function and avoids coupling to the
internal get_irq_status symbol.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: defaults
Review profile: CHILL
Plan: Pro
Run ID: 40028f94-5c5b-479d-b55a-cbaaaddb1333
📒 Files selected for processing (17)
.github/workflows/ci.ymlBdd/Targets/FreeRtosLwip/CMakeLists.txtBdd/Targets/FreeRtosLwip/FreeRTOSConfig.hBdd/Targets/FreeRtosLwip/README.mdBdd/Targets/FreeRtosLwip/lwipopts.hBdd/Targets/FreeRtosLwip/main.cBdd/Targets/FreeRtosLwip/netif/EthernetIf.cBdd/Targets/FreeRtosLwip/netif/EthernetIf.hBdd/Targets/FreeRtosLwip/netif/smsc9220/.clang-formatBdd/Targets/FreeRtosLwip/netif/smsc9220/smsc9220_emac_config.hBdd/Targets/FreeRtosLwip/netif/smsc9220/smsc9220_eth_drv.cBdd/Targets/FreeRtosLwip/netif/smsc9220/smsc9220_eth_drv.hBdd/Targets/FreeRtosLwip/solidsyslog_user_tunables.hCMakeLists.txtCMakePresets.jsonDEVLOG.mdci/docker-compose.bdd.yml
…sion, RX drain Apply the actionable, in-scope findings from the CodeRabbit review of #476, matching the established Bdd/Targets/FreeRtos (+TCP) sibling where it diverged: - main.c: check both xTaskCreate results and SemihostingExit(1) on failure rather than entering vTaskStartScheduler with a missing task (sibling does the same guard). - main.c: bump endpointVersion when `set host` changes the host, so a host change takes effect without also changing port (sibling already does this). - main.c: correct the LwipTcpipMarshal comment — tcpip_callback only blocks until the work is queued; synchronicity here comes from TCPIP_THREAD_PRIO being above every marshalling task. Records the LOCK_TCPIP_CORE follow-up. - EthernetIf.c: drain an oversize/corrupt RX frame from the FIFO in word-aligned chunks instead of returning NULL, which would re-peek the same frame forever and wedge receive. - README.md / DEVLOG.md: fix the vendored driver paths (netif/smsc9220/) and the tcpip_callback wording. Oracle re-run green (8 features / 25 scenarios / 96 steps, 0 failed). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
PR #476 review #4. tcpip_callback only blocks until the work is queued, so LwipTcpipMarshal previously relied on TCPIP_THREAD_PRIO preempting every marshalling task to appear synchronous — a correct-but-implicit invariant. LWIP_TCPIP_CORE_LOCKING is enabled, so run the callback in the caller's task context under a LOCK_TCPIP_CORE/UNLOCK_TCPIP_CORE pair instead: unconditionally synchronous regardless of priority, with no per-send mailbox message. The ARP warm-up query reuses the same marshal, which also closes the stack-`resolved` race in WarmUpGatewayArp. NetworkBringUp stays on tcpip_callback (its smsc9220_init calls vTaskDelay and must not run under the held core lock). Oracle re-run green (8 features / 25 scenarios / 96 steps, 0 failed). Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
Owner
Author
CodeRabbit review triageThanks @coderabbitai. Verified all 12 findings against the code and the merged Fixed in this PR
Oracle re-run green after each change: 8 features / 25 scenarios / 96 steps, 0 failed. Deferred (tracked, not dropped)
S28.09 stays UDP-scoped; the deferrals are real-hardware / project-wide hygiene that belong outside this story. |
Contributor
|
A few notes on the fixes:
The deferrals are sensibly scoped: All findings are either fixed or tracked — happy to resolve all outstanding review comments on this PR if you'd like. 🐇 ✓ ✏️ Learnings added
✅ Actions performedComments resolved. Approval is disabled; enable |
Contributor
☀️ Quality Summary 🚦 build-linux-gcc: 100% successful (✔️ 1344 passed) Created by Quality Monitor v1.14.0 (#f3859fd). More details are shown in the GitHub Checks Result. |
DavidCozens
added a commit
that referenced
this pull request
May 30, 2026
Adds `persist-credentials: false` to all 26 `actions/checkout` steps (25 in ci.yml, 1 in sbom.yml) so the checkout's GITHUB_TOKEN is not persisted in .git/config for later steps — the BDD compose lanes bind-mount the repo root into containers. Closes the zizmor `artipacked` finding deferred from #476. SHA-pins the three remaining tag-pinned actions so the workflow tree is zizmor `unpinned-uses`-clean: - dorny/test-reporter@v3 -> a43b3a5 # v3.0.0 - actions/cache@v4 -> (then bumped to v5, see below) - uhafner/quality-monitor@v1 -> 778e3a6 # v1.14.0 Sweeps the entire workflow tree (ci.yml, sbom.yml, release-please.yml) off the deprecated Node 20 runtime (forced off the runner from 2026-06-16) to the node24 release of each action: - actions/upload-artifact -> 043fb46 # v7.0.1 - actions/download-artifact -> 3e5f45b # v8.0.1 - actions/cache -> 27d5ce7 # v5.0.5 - actions/github-script -> ed59741 # v8.0.0 - actions/deploy-pages -> cd2ce8f # v5.0.0 - googleapis/release-please-action -> 45996ed # v5.0.0 Verified node24-clean: every pinned SHA's action.yml resolves to using:node24 (or composite/docker); no node20 and no unpinned uses remain. Behaviour notes (acceptable pre-release): download-artifact@v8 defaults a hash mismatch to a hard error; release-please-action@v5 marks node24 a breaking change but it is only the runtime bump. The deploy-pages and release-please lanes run on main only, so this PR's CI does not exercise them. Closes #477 Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
DavidCozens
added a commit
that referenced
this pull request
May 30, 2026
#485) Adds `persist-credentials: false` to all 26 `actions/checkout` steps (25 in ci.yml, 1 in sbom.yml) so the checkout's GITHUB_TOKEN is not persisted in .git/config for later steps — the BDD compose lanes bind-mount the repo root into containers. Closes the zizmor `artipacked` finding deferred from #476. SHA-pins the three remaining tag-pinned actions so the workflow tree is zizmor `unpinned-uses`-clean: - dorny/test-reporter@v3 -> a43b3a5 # v3.0.0 - actions/cache@v4 -> (then bumped to v5, see below) - uhafner/quality-monitor@v1 -> 778e3a6 # v1.14.0 Sweeps the entire workflow tree (ci.yml, sbom.yml, release-please.yml) off the deprecated Node 20 runtime (forced off the runner from 2026-06-16) to the node24 release of each action: - actions/upload-artifact -> 043fb46 # v7.0.1 - actions/download-artifact -> 3e5f45b # v8.0.1 - actions/cache -> 27d5ce7 # v5.0.5 - actions/github-script -> ed59741 # v8.0.0 - actions/deploy-pages -> cd2ce8f # v5.0.0 - googleapis/release-please-action -> 45996ed # v5.0.0 Verified node24-clean: every pinned SHA's action.yml resolves to using:node24 (or composite/docker); no node20 and no unpinned uses remain. Behaviour notes (acceptable pre-release): download-artifact@v8 defaults a hash mismatch to a hard error; release-please-action@v5 marks node24 a breaking change but it is only the runtime bump. The deploy-pages and release-please lanes run on main only, so this PR's CI does not exercise them. Closes #477 Co-authored-by: Claude Opus 4.8 (1M context) <noreply@anthropic.com>
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
1 participant
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.
Purpose
S28.09 (#472, epic E28) gives the
FreeRtosLwipBDD target a workingnetwork stack: a hand-written lwIP netif over Arm's smsc9220 LAN9118
driver, lwIP in
NO_SYS=0(tcpip-thread) mode, and a UDP loggingpipeline proven end-to-end against the syslog-ng oracle on QEMU.
Closes #472.
Change Description
Bdd/Targets/FreeRtosLwip/netif/smsc9220/(Apache-2.0 headers kept),isolated from
analyze-formatwith aDisableFormat.clang-format.Hand-written netif glue
netif/EthernetIf.c(RX task driven by IRQ-13EthernetISR, pbuf TX/RX,etharp_output).lwipopts.hruns the tcpip thread + contribsys_arch.c;FreeRTOSConfig.hmirrors the requirements; newsolidsyslog_user_tunables.h(MAX_MESSAGE_SIZE 512, FreeRtosMutex pool 2)flows in via the
freertos-cross-lwippreset.main.ccallstcpip_initpre-scheduler; netif bring-upruns from the interactive task via
tcpip_callback(smsc9220_init callsvTaskDelay, so it must run post-scheduler). lwIP-touching adapter opsare marshalled through a
tcpip_callback_with_blockshim installed withS28.06's
SolidSyslogLwipRaw_SetMarshal. UDP path = UdpSender +CircularBuffer + FreeRtosMutex + Service task + Meta/TimeQuality/Origin
SD. SwitchingSender TCP/TLS slots are wired to
NullSender(UDP onlythis story; TCP is S28.10).
LwipRawDatagramsends zero-copy(PBUF_REF over a transient buffer), so an ARP-miss first send had its
queued copy reference freed memory (seqId=1 lost, seqId=2 delivered).
Fix is
WarmUpGatewayArp:etharp_request(gw)at bring-up + polletharp_find_addr(marshalled off the tcpip thread) until resolvedbefore the pipeline goes live — keeps the zero-copy contract and pushes
the one-time resolution cost to setup. pcap now shows gratuitous ARP →
gw ARP request → reply → UDP seqId=1 → seqId=2 in order.
bdd-freertos-qemu-lwiplane +syslog-ng/behavecompose pair (UDP-only tag filter,
@freertoslwipwipper-scenarioescape hatch);
build-freertos-target-lwipuploads thebdd-tunables-freertos-lwipartifact. NOT insummary.needs—advisory until S28.11 promotes it.
Test Evidence
container):
docker compose -f ci/docker-compose.bdd.yml up --abort-on-container-exit --exit-code-from behave-freertos-lwip behave-freertos-lwip syslog-ng-freertos-lwip→ 8 features passed, 0 failed; 25 scenarios passed, 0 failed; 96 steps
passed, 0 failed (24 scenarios skipped — the TCP/TLS/mTLS/store
features excluded by the UDP-only tag filter). Delivery truth is the
syslog-ng oracle's "receives a message with priority/hostname/PROCID"
assertions, not an ad-hoc host listener (slirp does not reliably
deliver the NATed datagram to host loopback even though it is correct
on the wire).
arm-none-eabi-nmconfirmszero PlusTcp symbols linked into the lwIP ELF.
clang-format --dry-run --Werror(clang-format 19.1.7) clean.Areas Affected
Bdd/Targets/FreeRtosLwip/(Tier 3 best-effort BDD target) — new netif,driver, pipeline, tunables, presets.
.github/workflows/ci.yml,ci/docker-compose.bdd.yml— advisory lwIPlane + compose pair.
CMakeLists.txt/CMakePresets.json—freertos-cross-lwipwiring. No Tier-1/2 production source (
Core/,Platform/) touched.🤖 Generated with Claude Code
Summary by CodeRabbit
New Features
Bug Fixes
CI/Testing Improvements
Documentation