Misc fixes #393
Misc fixes #393
Conversation
|
Caution Review failedThe pull request is closed. 📝 WalkthroughWalkthroughAdds a new interface lifecycle event Pre-merge checks❌ Failed checks (1 warning, 1 inconclusive)
✅ Passed checks (1 passed)
📜 Recent review detailsConfiguration used: Path: .coderabbit.yaml Review profile: CHILL Plan: Pro 📒 Files selected for processing (11)
Tip 📝 Customizable high-level summaries are now available in beta!You can now customize how CodeRabbit generates the high-level summary in your pull requests — including its content, structure, tone, and formatting.
Example instruction:
Note: This feature is currently in beta for Pro-tier users, and pricing will be announced later. 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 |
christophefontaine
force-pushed
the
misc_fixes
branch
from
0935191 to
b186e0b
Compare
christophefontaine
force-pushed
the
misc_fixes
branch
from
b186e0b to
a48119c
Compare
christophefontaine
force-pushed
the
misc_fixes
branch
2 times, most recently
from
fc46f7e to
b36d252
Compare
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 2
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
modules/infra/control/loopback.c (1)
111-173: Fix mbuf leak on EAGAIN/EWOULDBLOCK in iface_loopback_pollWhen
read(lo->fd, ...)returns<= 0witherrno == EAGAIN || errno == EWOULDBLOCK, the functionreturns without freeingmbuf, leaking one mbuf per spurious wakeup. The newcp_rx_*updates are fine, but this branch should also free the buffer.A minimal fix is to reuse the existing
errpath:- if ((len = read(lo->fd, data, read_len)) <= 0) { - if (errno == EAGAIN || errno == EWOULDBLOCK) - return; + if ((len = read(lo->fd, data, read_len)) <= 0) { + if (errno == EAGAIN || errno == EWOULDBLOCK) + goto err;This keeps behavior identical while avoiding the leak.
🧹 Nitpick comments (2)
modules/infra/control/netlink.c (1)
234-243: Consider setting RT_SCOPE_LINK for link-local addresses.The /64 prefix for link-local addresses is correct per RFC 4291. However, line 244 sets
ifa_scope = RT_SCOPE_UNIVERSEfor all addresses. Link-local IPv6 addresses (fe80::/10) should typically useRT_SCOPE_LINKinstead ofRT_SCOPE_UNIVERSE.Consider this adjustment:
ifa = mnl_nlmsg_put_extra_header(nlh, sizeof(*ifa)); ifa->ifa_family = is_ipv4 ? AF_INET : AF_INET6; if (is_ipv4) { ifa->ifa_prefixlen = 32; + ifa->ifa_scope = RT_SCOPE_UNIVERSE; } else { // For Link local address, a /64 is required // to reach other nodes on the link - if (rte_ipv6_addr_is_linklocal(addr)) + if (rte_ipv6_addr_is_linklocal(addr)) { ifa->ifa_prefixlen = 64; - else + ifa->ifa_scope = RT_SCOPE_LINK; + } else { ifa->ifa_prefixlen = 128; + ifa->ifa_scope = RT_SCOPE_UNIVERSE; + } } - ifa->ifa_scope = RT_SCOPE_UNIVERSE; ifa->ifa_index = ifindex;modules/infra/cli/iface.c (1)
320-331: CP_ stats columns and row mapping are consistent; consider rates view*The four
CP_*columns are appended after the existing TX columns, and indices 7–10 correctly map tocp_rx_packets/bytesandcp_tx_packets/bytes, matching the existing%luformatting style.
If you want full symmetry later, you could also expose the CP_* counters iniface rates.Also applies to: 347-356
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (11)
frr/zebra_dplane_grout.c(5 hunks)modules/infra/api/gr_infra.h(2 hunks)modules/infra/api/iface.c(1 hunks)modules/infra/api/stats.c(3 hunks)modules/infra/cli/iface.c(4 hunks)modules/infra/control/gr_iface.h(1 hunks)modules/infra/control/iface.c(1 hunks)modules/infra/control/loopback.c(3 hunks)modules/infra/control/netlink.c(1 hunks)modules/ip6/control/nexthop.c(1 hunks)subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff(1 hunks)
🚧 Files skipped from review as they are similar to previous changes (2)
- frr/zebra_dplane_grout.c
- subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff
🧰 Additional context used
📓 Path-based instructions (1)
**/*.{c,h}
⚙️ CodeRabbit configuration file
**/*.{c,h}: -gr_vec_*()functions cannot fail. No need to check their return value.
gr_vec_free(x)always setsx = NULL. There is no risk of double free.ec_node_*()functions consume theirec_nodearguments. No leaks on error.rte_node->ctxis an uint8_t array of size 16, not a pointer.- Don't suggest to replace
assert()with graceful error checking.- We compile with
-std=gnu2x. Unnamed parameters in function definitions are valid.
Files:
modules/infra/control/netlink.cmodules/infra/api/iface.cmodules/infra/control/gr_iface.hmodules/infra/api/gr_infra.hmodules/infra/control/loopback.cmodules/infra/control/iface.cmodules/infra/api/stats.cmodules/ip6/control/nexthop.cmodules/infra/cli/iface.c
🧠 Learnings (9)
📓 Common learnings
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 372
File: smoke/cross_vrf_forward_test.sh:18-18
Timestamp: 2025-11-05T13:55:26.189Z
Learning: In the DPDK/grout codebase, VRF interfaces (named gr-vrf<id>) are automatically created when an interface is added to a non-existing VRF using port_add. The VRF creation is handled automatically by the event system in vrf_netlink.c, so no explicit VRF interface creation commands are needed in test scripts.
📚 Learning: 2025-11-05T14:28:28.817Z
Learnt from: rjarry
Repo: DPDK/grout PR: 372
File: modules/infra/datapath/loop_input.c:53-56
Timestamp: 2025-11-05T14:28:28.817Z
Learning: In modules/infra/datapath/loop_input.c, the l3_edges array is intentionally indexed using network byte order (rte_be16_t) for EtherType values. Both loopback_input_add_type() registration and loopback_input_process() lookup use network byte order consistently, so no byte order conversion is needed when accessing l3_edges.
Applied to files:
modules/infra/control/loopback.c
📚 Learning: 2025-11-05T13:55:26.189Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 372
File: smoke/cross_vrf_forward_test.sh:18-18
Timestamp: 2025-11-05T13:55:26.189Z
Learning: In the DPDK/grout codebase, VRF interfaces (named gr-vrf<id>) are automatically created when an interface is added to a non-existing VRF using port_add. The VRF creation is handled automatically by the event system in vrf_netlink.c, so no explicit VRF interface creation commands are needed in test scripts.
Applied to files:
modules/infra/control/iface.c
📚 Learning: 2025-09-25T07:52:17.403Z
Learnt from: rjarry
Repo: DPDK/grout PR: 312
File: frr/rt_grout.c:355-361
Timestamp: 2025-09-25T07:52:17.403Z
Learning: The nexthop_new() function in FRR's zebra codebase cannot fail and return NULL - it will abort() the process if memory allocation fails, so null checks after calling nexthop_new() are unnecessary.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-09-22T09:21:51.749Z
Learnt from: rjarry
Repo: DPDK/grout PR: 312
File: modules/ip/control/address.c:102-110
Timestamp: 2025-09-22T09:21:51.749Z
Learning: The rib4_insert() function in modules/ip/control/route.c takes ownership of the nexthop reference by calling nexthop_incref(nh) at the beginning and properly calls nexthop_decref(nh) on any failure paths via the fail: label, so callers don't need to manually decrement the reference count when rib4_insert fails.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-09-09T09:22:31.596Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 309
File: modules/srv6/datapath/srv6_local.c:97-101
Timestamp: 2025-09-09T09:22:31.596Z
Learning: In SRv6 IPv6 extension header parsing, when is_ipv6_ext[proto] is true, rte_ipv6_get_next_ext() will not return a negative value, making error checking unnecessary. An assert can be used as a defensive measure for future DPDK compatibility.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-09-22T09:21:51.749Z
Learnt from: rjarry
Repo: DPDK/grout PR: 312
File: modules/ip/control/address.c:102-110
Timestamp: 2025-09-22T09:21:51.749Z
Learning: The rib4_insert() function in the IP module takes ownership of the nexthop reference and automatically calls nexthop_decref(nh) on failure paths, so callers don't need to manually decrement the reference count when rib4_insert fails.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-09-09T06:02:47.703Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 309
File: modules/srv6/datapath/srv6_local.c:88-122
Timestamp: 2025-09-09T06:02:47.703Z
Learning: The DPDK function rte_ipv6_get_next_ext() only reads 2 bytes from the extension header, so a 2-byte precheck is sufficient before calling it.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-08-27T15:33:22.299Z
Learnt from: rjarry
Repo: DPDK/grout PR: 305
File: modules/ip6/control/route.c:407-413
Timestamp: 2025-08-27T15:33:22.299Z
Learning: DPDK rte_rib6_get_nxt() with RTE_RIB6_GET_NXT_ALL flag does not yield the default route ::/0 if configured. The explicit rte_rib6_lookup_exact() call for the default route is necessary to ensure complete route enumeration.
Applied to files:
modules/ip6/control/nexthop.c
🧬 Code graph analysis (4)
modules/infra/control/loopback.c (1)
modules/infra/control/gr_iface.h (2)
iface_get_stats(104-106)iface(16-22)
modules/infra/control/iface.c (2)
modules/infra/control/worker_test.c (1)
gr_event_push(37-37)modules/infra/control/gr_iface.h (1)
iface(16-22)
modules/infra/api/stats.c (1)
modules/infra/control/gr_iface.h (2)
iface_get_stats(104-106)iface(16-22)
modules/infra/cli/iface.c (1)
cli/table.c (1)
scols_line_sprintf(10-23)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (8)
- GitHub Check: build-and-tests (clang-16, none, debugoptimized, ubuntu-24.04, false)
- GitHub Check: build-and-tests (clang-18, none, debugoptimized, ubuntu-24.04, false)
- GitHub Check: build-and-tests (clang-15, none, debugoptimized, ubuntu-22.04, false)
- GitHub Check: build-and-tests (gcc-14, address, debug, ubuntu-24.04, -Dfrr=enabled, true)
- GitHub Check: build-and-tests (gcc-13, none, debugoptimized, ubuntu-24.04, -Dfrr=enabled, false)
- GitHub Check: build-cross-aarch64
- GitHub Check: rpm
- GitHub Check: deb
🔇 Additional comments (9)
modules/ip6/control/nexthop.c (1)
225-226: Good defensive NULL check.Prevents dereferencing
l3at line 228 when no L3 nexthop context exists. Ifl3is NULL, there are no held packets to flush, so jumping to cleanup is correct.modules/infra/control/netlink.c (1)
239-239: No issues found. API is available and type-safe.The
rte_ipv6_addr_is_linklocal()function is available in DPDK and is used throughout the codebase. Theconst void *parameter at line 239 is type-safe because the runtime length check (line 213:addr_len == sizeof(ip4_addr_t)) guarantees that when the IPv6 branch executes, the pointer references a valid IPv6 address. The implicit conversion toconst struct rte_ipv6_addr *is safe.modules/infra/api/iface.c (1)
164-175: Serializer ev_count matches new IFACE_ADD event set
ev_count = 6matches the six entries inev_types, andGR_EVENT_IFACE_ADDis correctly added as the first iface event, so the serializer stays consistent with the updated enum.modules/infra/control/iface.c (1)
130-137: ADD event emitted before POST_ADD on fully initialized ifacePushing
GR_EVENT_IFACE_ADDright after installing the iface and zeroingiface_stats, and beforeGR_EVENT_IFACE_POST_ADD, is consistent with a “create” then “post‑add” phase for consumers.modules/infra/cli/iface.c (1)
545-555: CLI event printer now covers GR_EVENT_IFACE_ADD cleanly
iface_event_printhandlesGR_EVENT_IFACE_ADD("add") andGR_EVENT_IFACE_POST_ADD("post add"), andprinter.ev_typesplusev_count = 6are kept in sync, so CLI event output will see both phases without mismatch.Also applies to: 579-587
modules/infra/control/loopback.c (1)
53-109: CP TX counters integrated into loopback_txUsing
iface_get_stats(rte_lcore_id(), d->iface->id)and bumpingcp_tx_packets/cp_tx_bytesafter preparing the write keeps control‑plane TX accounting consistent with the per‑lcoreiface_statslayout.modules/infra/control/gr_iface.h (1)
90-99: iface_stats extended with cp_ counters; initialization still correct*Adding
cp_rx_packets/bytesandcp_tx_packets/bytesafter the existing fields preserves the struct layout, and thememset(iface_stats[ifid], ...)iniface_createwill also zero these new counters, so readers aggregating stats remain consistent.modules/infra/api/stats.c (2)
269-284: LGTM: Control-plane counter aggregation is correct.The initialization and aggregation of
cp_*counters follows the same pattern as existingrx_*andtx_*counters. The per-core stats are correctly summed into per-interface totals.
397-416: LGTM: Telemetry output correctly includes control-plane counters.The control-plane counters are properly accumulated across cores and exposed in the telemetry data alongside existing metrics.
christophefontaine
force-pushed
the
misc_fixes
branch
5 times, most recently
from
cdb8553 to
202346b
Compare
There was a problem hiding this comment.
Actionable comments posted: 0
Caution
Some comments are outside the diff and can’t be posted inline due to platform limitations.
⚠️ Outside diff range comments (1)
modules/ip6/control/nexthop.c (1)
102-117: LOG uses IPv4 formatter and wrong pointer type for IPv6 destination
LOG(DEBUG, IP4_F " hold queue full", &dst);looks wrong in two ways:
- Uses
IP4_Fwhiledstis an IPv6 address.- Passes
&dst(pointer to pointer) instead ofdst(pointer to the IPv6 address).This likely prints garbage and may read unintended memory. Consider:
- LOG(DEBUG, IP4_F " hold queue full", &dst); + LOG(DEBUG, IP6_F " hold queue full", dst);
🧹 Nitpick comments (1)
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff (1)
20-35: Link‑local generation now matches RFC 4291 modified EUI‑64Switching the reference to RFC 4291 §2.5.1 and XORing the first IID byte with
0x02fixes the previous non‑compliance and aligns the implementation with the standard modified EUI‑64 mapping from Ethernet MACs.If you want to make the bit‑twiddle more self‑documenting, you could optionally add a short comment:
- ip->a[8] = mac->addr_bytes[0]; + /* Flip the U/L bit per RFC 4291, section 2.5.1. */ + ip->a[8] = mac->addr_bytes[0] ^ 0x02;
📜 Review details
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (18)
Containerfile.frr(1 hunks)debian/control(1 hunks)frr/zebra_dplane_grout.c(5 hunks)main/api.c(1 hunks)meson.build(1 hunks)modules/infra/api/gr_infra.h(2 hunks)modules/infra/api/iface.c(1 hunks)modules/infra/api/stats.c(3 hunks)modules/infra/cli/iface.c(4 hunks)modules/infra/control/gr_iface.h(1 hunks)modules/infra/control/iface.c(1 hunks)modules/infra/control/loopback.c(4 hunks)modules/infra/control/netlink.c(1 hunks)modules/ip6/control/nexthop.c(1 hunks)subprojects/dpdk.wrap(1 hunks)subprojects/frr.wrap(1 hunks)subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff(1 hunks)subprojects/packagefiles/frr/meson-add-dependency-definition.patch(1 hunks)
✅ Files skipped from review due to trivial changes (1)
- subprojects/frr.wrap
🚧 Files skipped from review as they are similar to previous changes (5)
- modules/infra/control/iface.c
- modules/infra/api/gr_infra.h
- modules/infra/control/loopback.c
- modules/infra/control/netlink.c
- modules/infra/api/iface.c
🧰 Additional context used
📓 Path-based instructions (1)
**/*.{c,h}
⚙️ CodeRabbit configuration file
**/*.{c,h}: -gr_vec_*()functions cannot fail. No need to check their return value.
gr_vec_free(x)always setsx = NULL. There is no risk of double free.ec_node_*()functions consume theirec_nodearguments. No leaks on error.rte_node->ctxis an uint8_t array of size 16, not a pointer.- Don't suggest to replace
assert()with graceful error checking.- We compile with
-std=gnu2x. Unnamed parameters in function definitions are valid.
Files:
frr/zebra_dplane_grout.cmodules/infra/cli/iface.cmodules/infra/control/gr_iface.hmain/api.cmodules/infra/api/stats.cmodules/ip6/control/nexthop.c
🧠 Learnings (9)
📓 Common learnings
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 372
File: smoke/cross_vrf_forward_test.sh:18-18
Timestamp: 2025-11-05T13:55:26.189Z
Learning: In the DPDK/grout codebase, VRF interfaces (named gr-vrf<id>) are automatically created when an interface is added to a non-existing VRF using port_add. The VRF creation is handled automatically by the event system in vrf_netlink.c, so no explicit VRF interface creation commands are needed in test scripts.
📚 Learning: 2025-11-05T13:55:26.189Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 372
File: smoke/cross_vrf_forward_test.sh:18-18
Timestamp: 2025-11-05T13:55:26.189Z
Learning: In the DPDK/grout codebase, VRF interfaces (named gr-vrf<id>) are automatically created when an interface is added to a non-existing VRF using port_add. The VRF creation is handled automatically by the event system in vrf_netlink.c, so no explicit VRF interface creation commands are needed in test scripts.
Applied to files:
frr/zebra_dplane_grout.c
📚 Learning: 2025-09-05T07:06:51.554Z
Learnt from: rjarry
Repo: DPDK/grout PR: 294
File: modules/policy/cli/meson.build:4-8
Timestamp: 2025-09-05T07:06:51.554Z
Learning: In this project, cli_src is a global variable that gets populated by individual modules (like modules/policy/cli/meson.build) and is ultimately consumed by an executable() call in the top-level meson.build file. This creates a modular CLI build where each module contributes its CLI sources to the main CLI binary.
Applied to files:
subprojects/packagefiles/frr/meson-add-dependency-definition.patch
📚 Learning: 2025-09-09T09:22:31.596Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 309
File: modules/srv6/datapath/srv6_local.c:97-101
Timestamp: 2025-09-09T09:22:31.596Z
Learning: In SRv6 IPv6 extension header parsing, when is_ipv6_ext[proto] is true, rte_ipv6_get_next_ext() will not return a negative value, making error checking unnecessary. An assert can be used as a defensive measure for future DPDK compatibility.
Applied to files:
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diffmodules/ip6/control/nexthop.c
📚 Learning: 2025-09-09T06:02:47.703Z
Learnt from: maxime-leroy
Repo: DPDK/grout PR: 309
File: modules/srv6/datapath/srv6_local.c:88-122
Timestamp: 2025-09-09T06:02:47.703Z
Learning: The DPDK function rte_ipv6_get_next_ext() only reads 2 bytes from the extension header, so a 2-byte precheck is sufficient before calling it.
Applied to files:
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff
📚 Learning: 2025-08-27T15:33:22.299Z
Learnt from: rjarry
Repo: DPDK/grout PR: 305
File: modules/ip6/control/route.c:407-413
Timestamp: 2025-08-27T15:33:22.299Z
Learning: DPDK rte_rib6_get_nxt() with RTE_RIB6_GET_NXT_ALL flag does not yield the default route ::/0 if configured. The explicit rte_rib6_lookup_exact() call for the default route is necessary to ensure complete route enumeration.
Applied to files:
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff
📚 Learning: 2025-11-05T14:28:28.817Z
Learnt from: rjarry
Repo: DPDK/grout PR: 372
File: modules/infra/datapath/loop_input.c:53-56
Timestamp: 2025-11-05T14:28:28.817Z
Learning: In modules/infra/datapath/loop_input.c, the l3_edges array is intentionally indexed using network byte order (rte_be16_t) for EtherType values. Both loopback_input_add_type() registration and loopback_input_process() lookup use network byte order consistently, so no byte order conversion is needed when accessing l3_edges.
Applied to files:
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff
📚 Learning: 2025-09-25T07:52:17.403Z
Learnt from: rjarry
Repo: DPDK/grout PR: 312
File: frr/rt_grout.c:355-361
Timestamp: 2025-09-25T07:52:17.403Z
Learning: The nexthop_new() function in FRR's zebra codebase cannot fail and return NULL - it will abort() the process if memory allocation fails, so null checks after calling nexthop_new() are unnecessary.
Applied to files:
modules/ip6/control/nexthop.c
📚 Learning: 2025-09-22T09:21:51.749Z
Learnt from: rjarry
Repo: DPDK/grout PR: 312
File: modules/ip/control/address.c:102-110
Timestamp: 2025-09-22T09:21:51.749Z
Learning: The rib4_insert() function in modules/ip/control/route.c takes ownership of the nexthop reference by calling nexthop_incref(nh) at the beginning and properly calls nexthop_decref(nh) on any failure paths via the fail: label, so callers don't need to manually decrement the reference count when rib4_insert fails.
Applied to files:
modules/ip6/control/nexthop.c
🧬 Code graph analysis (2)
modules/infra/cli/iface.c (1)
cli/table.c (1)
scols_line_sprintf(10-23)
modules/infra/api/stats.c (1)
modules/infra/control/gr_iface.h (2)
iface_get_stats(102-104)iface(16-22)
⏰ Context from checks skipped due to timeout of 90000ms. You can increase the timeout in your CodeRabbit configuration to a maximum of 15 minutes (900000ms). (5)
- GitHub Check: build-and-tests (clang-18, none, debugoptimized, ubuntu-24.04, false)
- GitHub Check: build-and-tests (gcc-13, none, debugoptimized, ubuntu-24.04, -Dfrr=enabled, false)
- GitHub Check: build-and-tests (clang-16, none, debugoptimized, ubuntu-24.04, false)
- GitHub Check: build-and-tests (clang-15, none, debugoptimized, ubuntu-22.04, false)
- GitHub Check: build-and-tests (gcc-14, address, debug, ubuntu-24.04, -Dfrr=enabled, true)
🔇 Additional comments (19)
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diff (1)
7-13: Test expectation correctly updated for modified EUI‑64 IIDThe new expected address (
0x067bvs0x047b) matches the^ 0x02flip of the MAC’s first byte, so the unit test now accurately reflects the updated link‑local construction.debian/control (1)
99-99: FRR version bump is consistent.The version update to 10.5.0 aligns with changes in other packaging files and the FRR patch.
Containerfile.frr (1)
16-16: No issues found. URL is accessible.The frr-10.5.0 tag exists and the docker-start script is available. The HTTP 200 response confirms the resource is accessible and the line is correct.
subprojects/dpdk.wrap (1)
9-10: Patch file verified and present.The patch file
subprojects/packagefiles/dpdk/net-ipv6-link-local-compliance-with-rfc-4291.diffexists at the expected location and contains valid diff content for IPv6 link-local RFC 4291 compliance updates.meson.build (1)
68-68: Confirm platform=generic is intentional; it prioritizes portability over per-CPU optimization.The
platform=genericoption in DPDK 24.11.1 builds with a minimal CPU/instruction-set baseline (e.g., corei7 on x86) for broad compatibility across machines, rather than machine-specific optimizations. Verify this aligns with your deployment requirements and performance expectations.subprojects/packagefiles/frr/meson-add-dependency-definition.patch (1)
64-64: FRR 10.5.0 is officially released and available.FRRouting 10.5.0 was officially released (Nov 9–10, 2025) and is available for download from packages, snaps, Docker, and GitHub release. The version specification is valid.
main/api.c (1)
65-65: LGTM!Documentation updated to reflect the new event enum.
frr/zebra_dplane_grout.c (4)
240-240: LGTM!Subscription correctly updated to the new
GR_EVENT_IFACE_ADDevent.
318-319: LGTM!String mappings added for new request types.
Also applies to: 332-333
380-381: LGTM!String mapping added for the new interface addition event.
438-438: LGTM!Event handling correctly updated to process the new addition event.
modules/infra/api/stats.c (3)
269-272: LGTM!Control-plane counters properly initialized.
281-284: LGTM!Per-core control-plane counters correctly aggregated into interface totals.
397-397: LGTM!Control-plane counters properly exposed in telemetry output.
Also applies to: 404-407, 413-416
modules/infra/cli/iface.c (3)
316-319: LGTM!Control-plane counter columns properly defined with right alignment.
342-345: LGTM!Control-plane counter data correctly populated in table output.
539-544: LGTM!Event printer updated to handle both the new
GR_EVENT_IFACE_ADDand existingGR_EVENT_IFACE_POST_ADDevents.Also applies to: 568-570
modules/infra/control/gr_iface.h (1)
93-96: LGTM!Control-plane counter fields properly added to the interface statistics structure.
modules/ip6/control/nexthop.c (1)
225-227: NULL-guard before flushing held packets avoids NDP crashAdding
if (l3 == NULL) goto free;here is correct and prevents a null dereference when the probe doesn’t resolve to an L3 nexthop (e.g. unspecified/mcast sender, DAD). Control flow remains consistent with other error paths in this function.
christophefontaine
force-pushed
the
misc_fixes
branch
3 times, most recently
from
c0575cc to
adb556f
Compare
christophefontaine
force-pushed
the
misc_fixes
branch
from
adb556f to
6c08644
Compare
christophefontaine
force-pushed
the
misc_fixes
branch
from
6c08644 to
adc8203
Compare
Fix a race condition where we get a packet on interface creation before every thing is properly set up. Fixes: 35a0677 ("ndp: handle nexthop updates in control plane") Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
Current synchronisation pushes a /128 for all ip6 addresses. Yet, a /64 is required for link local addresses, so OSPF6d can reach its neighbors. Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
Add new fields in the existing stats structure to track the number of packets sent/received by the control plane interfaces, gr-loop included. Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
We register in grout to the event "GR_EVENT_IFACE_ADD_POST" to add a link local ipv6 address. Yet, this action also trigger an event "GR_EVENT_IP6_ADDR_ADD", with an iface id which doesn't exists yet. In addition to GR_EVENT_IFACE_ADD_POST, we introduce a new event "GR_EVENT_IFACE_ADD", which will be pushed before "ADD_POST". grout# events show > iface add: gr-loop0 type=loopback id=256 vrf=0 mtu=1500 > iface post add: gr-loop0 type=loopback id=256 vrf=0 mtu=1500 > iface add: p0 type=port id=257 vrf=0 mtu=1500 > route6 add: vrf=0 fe80::7077:78ff:fe2f:c803/64 origin=link via type=L3 iface=257 vrf=0 origin=INTERNAL af=IPv6 addr=fe80::7077:78ff:fe2f:c803/64 mac=72:77:78:2f:c8:03 static local link > addr6 add: iface=257 fe80::7077:78ff:fe2f:c803/64 > iface post add: p0 type=port id=257 vrf=0 mtu=1500 > iface up: p0 type=port id=257 vrf=0 mtu=1500 Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
On interface creation, use the new GR_EVENT_IFACE_ADD instead of GR_EVENT_IFACE_POST_ADD to create the interface. Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
Add missing string in the debug function 'gr_req_type_to_str'. Fixes: 5e970e0 ("frr: use tunsrc api") Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
Fix mbuf leak on EAGAIN/EWOULDBLOCK in iface_loopback_poll. Fixes: d0de3b2 ("loopback: add loopback interface") Reported-by: coderabbitai Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
Fix debug message using IP4_F for an IPv6 address. Fixes: 9cb61fd ("ip6_output: handle unresolved nexthops in control plane") Reported-by: coderabbitai Signed-off-by: Christophe Fontaine <[email protected]> Reviewed-by: Robin Jarry <[email protected]>
grout-bot
force-pushed
the
misc_fixes
branch
from
adc8203 to
c3e60ee
Compare
3 participants
Summary by CodeRabbit
New Features
Bug Fixes
Other