News › networking
By The Linux Camp News, October 10, 2026
Maintainers deferred less pressing networking fixes while retaining regression repairs for Linux 7.3. The confirmed milestone is a merged pull request, not a released 7.3-rc7 kernel.
Linux networking maintainer Jakub Kicinski reported on October 8, 2026 that maintainers had diverted 50 fixes to net-next and asked subsystem maintainers to substantially trim their submissions for Linux 7.3-rc7. The remaining pull request was confirmed merged on October 9 at 04:47 UTC by the kernel pull-request tracker bot. As of the October 10 source retrieval, the confirmed status was merged code; Phoronix described the release candidate itself as due Sunday. 3 2 1
This is release-queue triage rather than one networking algorithm change. Net-next is the queue for networking work intended for the next kernel cycle. Phoronix reports that less pressing fixes began moving there late in the 7.3 cycle and identifies Linux 7.4 as their destination. Deferring them reduces the changes entering the late release candidate without discarding the fixes. 1
The deferral is selective, not a networking-fix freeze. The retained batch covers wireless, WireGuard, CAN and Bluetooth. It repairs a current-cycle Bluetooth RFCOMM deadlock alongside older regressions affecting iavf reconfiguration with bonding and mana packet-forwarding performance. It also removes buggy stmmac VLAN acceleration support. Those driver and workflow distinctions matter more to an operator than the aggregate count of deferred patches. 3
For engineers tracking a specific fault, the practical inference is to check whether its fix appears in the merged batch or was routed to net-next before planning release-candidate testing. A blanket assumption that networking repairs must wait for 7.4 would miss the regression fixes retained for 7.3. 1 3
Kicinski also identified one VLAN-handling regression still awaiting resolution when he submitted the request. The subsequent merge confirmation establishes that the submitted batch landed, not that this outstanding problem was resolved. The announcement therefore cannot serve as an all-clear for VLAN workloads. 3 2
The broader explanation that AI-assisted reports and patches are driving submission pressure comes from Phoronix. Kicinski's announcement calls the nxpwifi changes “LLM-ish,” but that description is not a provenance count. It does not establish that all 50 deferred fixes were AI-generated or quantify how much review work AI assistance caused. 1 3