[24.10] hev-socks5-tunnel: allow multiple instances

[Ser9ei]

Add support for running multiple service instances.

• switch UCI config from a single named section to "instance" sections
• update the procd init script to start multiple instances
• migrate existing single-instance configurations during package upgrade (uci-defaults script)
• bump PKG_RELEASE

(cherry picked from commit d04b36b)

📦 Package Details

Maintainer: @heiher
_{(You can find this by checking the history of the package Makefile.)}

Description:

---

🧪 Run Testing Details

• OpenWrt Version: 25.12.5
• OpenWrt Target/Subtarget: mediatek/filogic
• OpenWrt Device: Xiaomi Mi Router AX3000T

[openwrt-ai]

Backport review — the diff itself is a verbatim match for the change on master (I diffed this PR's patch against the upstream commit and the resulting package trees are byte-identical, and the 24.10 base tree for net/hev-socks5-tunnel/ was already identical to the upstream commit's parent). No deviations introduced by the cherry-pick. One trailer issue below.

Commit checks

• 9ab699d2 "hev-socks5-tunnel: allow multiple instances" — the (cherry picked from commit f9fdfcf09a3a79251ccd88853cbda6331fe43722) trailer points at a commit that is not on master. f9fdfcf0 is the head commit of your own PR branch in #30143 (author and committer both Sergei S., committed 2026-08-07). The commit that actually landed on master is d04b36bbba607e65072ecd0abfae5a89af468607 (committed by Alexandru Ardelean, 2026-08-09) — same tree, same message, different SHA because it was re-applied on merge. Please re-do the pick with git cherry-pick -x d04b36bbba607e65072ecd0abfae5a89af468607 (or just fix the trailer) so the backport references a SHA that exists on master. The same applies to the sibling backport #30272, which carries the identical trailer.

Nothing else blocks a merge from my side; the code change is unmodified from what was already reviewed and merged on master.

---

Generated by Claude Code

[openwrt-ai]

nit: the PR body's Run Testing Details says OpenWrt Version: 25.12.5, but this PR targets openwrt-24.10. The same body text appears on the sibling backport #30272 ([25.12]), so it looks copy-pasted rather than describing a 24.10 run. Was the migration path (old config hev-socks5-tunnel 'config' section → instance sections via the new uci-defaults script) also exercised on a 24.10 build? Not a blocker, just worth correcting the body so the record is accurate.

---

Generated by Claude Code

[Ser9ei]

The CI failure appears to be unrelated to this PR. One of the CI tests fails while downloading a package index from the OpenWrt download server.
The same failure is reproducible on other PRs, while the other CI jobs pass.
No changes to this PR are required.

[GeorgeSapkin]

I don't think it makes sense backporting features and/or breaking changes to an almost EoL ¹ branch.

Footnotes

1. https://openwrt.org/releases/24.10/notes-24.10.8 ↩

[Ser9ei]

@GeorgeSapkin I'm not opposed to closing this cherry-pick PR for the 24.10 branch.

Just for context: the package was recently (4 days ago) updated to the new upstream release (2.17.0) in openwrt-24.10: 412c6ae

@heiher, could you please share your feedback?

[heiher]

@heiher, could you please share your feedback?

Thanks! There are still users on 24.10, so I'd prefer to backport this change if possible. This should also help minimize conflicts when we backport future changes to old releases.

[openwrt-ai]

Reviewed 1 new commit; no new issues found.

The trailer issue from my previous review is resolved: 8316be9a now carries (cherry picked from commit d04b36bbba607e65072ecd0abfae5a89af468607), and d04b36bb is on master. The tree is unchanged from the version I reviewed at 9ab699d2 (git diff 9ab699d2..8316be9a is empty) — only the commit message was amended.

Re-verified the backport against upstream: all four files (Makefile, files/hev-socks5-tunnel.init, files/hev-socks5-tunnel.config, files/90-hev-socks5-tunnel) are byte-identical to their post-state in d04b36bb. The branch also sits on top of 412c6ae4 "hev-socks5-tunnel: update to 2.17.0" on openwrt-24.10, so the PKG_RELEASE bump 1 → 2 at 2.17.0 lines up with what master ships. No deviations introduced by the cherry-pick.

The red Feeds Package Test Build jobs are not caused by this change — opkg update inside the test container fails on a missing kmods index:

Downloading https://downloads.openwrt.org/releases/24.10-SNAPSHOT/targets/malta/be/kmods/6.6.119-1-4880f58f355f8d52fbb2985e0e649b73/Packages.gz
*** Failed to download the package list from .../kmods/.../Packages.gz
 * opkg_download: Failed to download ..., wget returned 8.

Same failure across all failing arches, matching @Ser9ei's assessment in the thread.

The remaining open question is the policy one raised by @GeorgeSapkin (backporting a feature to a near-EoL branch) versus @heiher's preference to take it — that's a maintainer call, not something I can settle. Nothing in the diff blocks a merge.

---

Generated by Claude Code