kernel: backport RDMA/core route-entry loopback detection

[AZaugg]

Merge Checklist

All boxes should be checked before merging the PR (just tick any boxes which don't apply to this PR)

• The toolchain has been rebuilt successfully (or no changes were made to it)
• The toolchain/worker package manifests are up-to-date
• Any updated packages successfully build (or no packages were changed)
• Packages depending on static components modified in this PR (Golang, *-static subpackages, etc.) have had their Release tag incremented.
• Package tests (%check section) have been verified with RUN_CHECK=y for existing SPEC files, or added to new SPEC files
• All package sources are available
• cgmanifest files are up-to-date and sorted (./cgmanifest.json, ./toolkit/scripts/toolchain/cgmanifest.json, .github/workflows/cgmanifest.json)
• LICENSE-MAP files are up-to-date (./LICENSES-AND-NOTICES/SPECS/data/licenses.json, ./LICENSES-AND-NOTICES/SPECS/LICENSES-MAP.md, ./LICENSES-AND-NOTICES/SPECS/LICENSE-EXCEPTIONS.PHOTON)
• All source files have up-to-date hashes in the *.signatures.json files
• sudo make go-tidy-all and sudo make go-test-coverage pass
• Documentation has been updated to match any changes to the build system
• Ready to merge

---

Summary

Backport upstream commit c31e4038c97f ("RDMA/core: Use route entry flag to decide on loopback traffic") to the 6.6.139.1 AZL3 kernel.

In multi-NIC RoCE setups where the source and destination IPs live on netdevs enslaved to a VRF, addr_resolve() picks the VRF as the next-hop device. Because the VRF is not IFF_LOOPBACK, the existing loopback detection misses the local route and rdma_set_src_addr_rcu() leaves the RDMA destination MAC pointing at the VRF netdevice, causing ib_write_bw to time out. The upstream fix replaces the IFF_LOOPBACK check with a route-type (RTN_LOCAL / RTF_LOCAL) check so loopback is detected correctly with or without VRF.

Adjustments for 6.6:

• dst_rtable() is not present in v6.6 yet; replaced with a direct (const struct rtable *) cast (struct dst_entry is the first field of struct rtable, so it is equivalent to the upstream container_of).
• Verified that the patch applies cleanly to both v6.6.121 and v6.6.139 (drivers/infiniband/core/addr.c is byte-identical between them).

Spec/manifest changes (Release 1 -> 2, matching the BBR3 / IKCONFIG_PROC companion-bump pattern):

• SPECS/kernel/kernel.spec: add Patch1, bump Release
• SPECS/kernel/kernel-uki.spec: bump Release to match kernel
• SPECS/kernel-64k/kernel-64k.spec: bump Release to match kernel
• SPECS/kernel-headers/kernel-headers.spec: bump Release to match kernel
• SPECS-EXTENDED/kernel-ipe/kernel-ipe.spec: bump Release to match kernel
• SPECS-SIGNED/kernel-signed/kernel-signed.spec: bump Release to match kernel
• SPECS-SIGNED/kernel-64k-signed/kernel-64k-signed.spec: bump Release to match kernel
• SPECS-SIGNED/kernel-uki-signed/kernel-uki-signed.spec: bump Release to match kernel
• toolkit/resources/manifests/package/{pkggen_core,toolchain}_{x86_64,aarch64}.txt: bump kernel-headers / kernel-cross-headers to -2

Change Log

• Change
• Change
• Change

Does this affect the toolchain?

YES

Associated issues

• #xxxx

Links to CVEs

• https://nvd.nist.gov/vuln/detail/CVE-YYYY-XXXX

Test Methodology

• Pipeline build id: xxxx