Hello:
This series was applied to netdev/net-next.git (main)
by Jakub Kicinski [off-list ref]:
On Fri, 6 Nov 2026 19:14:40 +0200 you wrote:
Fix false bind() exhaustion when an odd-sized local port range still
contains an eligible free port. The wrap can flip parity in odd-sized
ranges; even-sized scan sequences are unchanged. The namespace-wide
connect() allocator is not changed by this series.
Earlier x86-64 KVM validation of this unchanged allocator/test code:
- Regression test: four failures before, zero after (eight cases).
- Full port-range suite: four failures before, zero after (80 cases);
two existing SCTP XPASS results unchanged.
- Go net.Listen, 100 attempts per IP version: 100 failures before,
zero after; explicit free-port controls succeed in both kernels.
[...]
Here is the summary with links:
- [net-next,v2,1/2] tcp: avoid false bind exhaustion with odd-sized port ranges
https://git.kernel.org/netdev/net-next/c/c7d11f97e8a2
- [net-next,v2,2/2] selftests/net: cover odd-sized local port ranges with one free port
https://git.kernel.org/netdev/net-next/c/ca133dd5fb20
You are awesome, thank you!
--
Deet-doot-dot, I am a bot.
https://korg.docs.kernel.org/patchwork/pwbot.html