Fix build #54: FINAL umount -> sys_umount fix for kernel 4.9

[Crybyte]

Automated fix for crDroid++ kernel build failure.

Issue: Build #53 failed with undefined reference to 'umount' at line 566

Root Cause:

• The pre-patched core_hook.c.patched file is applied AFTER v38 fix scripts run
• This overwrites any umount() -> sys_umount() replacements made by v38
• The resulting file still calls umount() which doesn't exist in kernel 4.9

Fix:

• Added fix_kernelsu_49_v39.py with aggressive umount() -> sys_umount() replacement
• Updated workflow to run v39 AFTER the pre-patched file is applied
• v39 ensures ALL umount() calls are replaced with sys_umount(path, 0)
• Also adds sys_umount declaration for kernel 4.9 compatibility

Changes:

• New file: build-scripts/fix_kernelsu_49_v39.py (aggressive regex-based replacement)
• Modified: .github/workflows/crdroid-plus-kernel-build.yml (runs v39 after pre-patched file)

Related: Build #54 (attempt #54 overall, #53 failed with umount error)

cc: @coderabbitai

Summary by CodeRabbit

• Chores
  • Added an extra automated fix step to the kernel build pipeline that runs a maintenance utility during builds.
  • Introduced a standalone CLI maintenance script that locates and updates affected kernel source files, applies targeted textual fixes, ensures required declarations exist, and prints a per-file status summary.

[coderabbitai]

Caution

Review failed

Pull request was closed or merged during review

📝 Walkthrough

Walkthrough

Adds a CI workflow step and a new Python fixer that scans kernel source for drivers/kernelsu/core_hook.c, aggressively rewrites umount(...) to sys_umount(..., 0), inserts a sys_umount declaration if missing, and the workflow includes a repo-root fallback that alters the kernel-dir path used for the script.

Changes

Cohort / File(s)	Summary
Workflow update ​.github/workflows/crdroid-plus-kernel-build.yml	Inserted "CRITICAL FIX v39" step that runs fix_kernelsu_49_v39.py preferentially from build-scripts/, with a fallback to the repo root; fallback uses a different kernel-dir path variant (~/kernel_build/... vs ~/kernel-build/...) when invoking the script.
KernelSU v39 fixer script build-scripts/fix_kernelsu_49_v39.py	New CLI Python script that finds core_hook.c files in a kernel dir, de-duplicates by realpath, performs aggressive regex replacements changing unprefixed umount(X) to sys_umount(X, 0), reports remaining umount( occurrences, and inserts an asmlinkage long sys_umount(const char __user *name, int flags); declaration after the last #include if absent.

Estimated code review effort

🎯 3 (Moderate) | ⏱️ ~20 minutes

Possibly related PRs

• Fix build #50: v36 umount fix for kernel 4.9 compatibility #28: Implements a fixer that edits drivers/kernelsu/core_hook.c with the same umount→sys_umount rewrite and declaration insertion.
• Fix build #52: Replace umount() with sys_umount() for kernel 4.9 #30: Adds a workflow and Python fixer targeting the identical CoreHook umount-to-sys_umount changes.
• Fix build #49: Replace umount with sys_umount for kernel 4.9 #27: Introduces a similar build step and script that perform aggressive umount replacements and ensure sys_umount is declared.

Poem

🐰 I hopped through code and found a trace,
umounts reborn with a sys_ embrace,
a declaration snug after includes' line,
v39 marched in—now the kernel will shine! ✨

🚥 Pre-merge checks | ✅ 2 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 75.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (2 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The title directly and clearly summarizes the main change: a final fix for umount to sys_umount conversion for kernel 4.9 compatibility, which aligns with the primary objective of resolving Build #54.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch fix/build-54-umount-final

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 3

🤖 Prompt for all review comments with AI agents

Verify each finding against the current code and only fix it if needed.

Inline comments:
In @.github/workflows/crdroid-plus-kernel-build.yml:
- Around line 447-450: The fallback invocation passes the wrong path string
(uses ~/kernel_build with an underscore) which will fail to locate the kernel
tree; update the command that runs fix_kernelsu_49_v39.py so the directory
argument and KERNEL_DIR environment value use the correct hyphenated path
(~/kernel-build/android_kernel_oneplus_sdm845) to match the rest of the workflow
and ensure core_hook.c files are found.

In `@build-scripts/fix_kernelsu_49_v39.py`:
- Around line 89-119: The function process_core_hook_files currently iterates
possible_paths and may process the same real file multiple times; change it to
deduplicate by tracking resolved real paths (e.g., a seen set) before processing
each entry: after computing real_path, skip the iteration if real_path is
already in seen, otherwise add real_path to seen and proceed with
aggressive_umount_fix and ensure_sys_umount_declaration, ensuring results only
contain one tuple per real_path; reference process_core_hook_files,
possible_paths, real_path, results, aggressive_umount_fix and
ensure_sys_umount_declaration when making the change.
- Around line 147-151: The loop unpacks (path, fixed, declared) from results but
never uses declared; update the summary inside the for loop (the for path,
fixed, declared in results block) to include declaration info — e.g. compute
status = "FIXED" if fixed else "OK" and append ", DECLARED" when declared is
truthy (or similar like status += " (declared)" ), then print f"  [{status}]
{path}"; alternatively if you intentionally discard declared, rename the third
unpack to _declared to signal it's unused (change for path, fixed, declared in
results to for path, fixed, _declared in results).

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Organization UI

Review profile: ASSERTIVE

Plan: Pro

Run ID: def65740-7967-46cb-81ec-f861a5781b42

📥 Commits

Reviewing files that changed from the base of the PR and between c2c389e and f59dd7e.

📒 Files selected for processing (2)

• .github/workflows/crdroid-plus-kernel-build.yml
• build-scripts/fix_kernelsu_49_v39.py

📜 Review details

🧰 Additional context used

🪛 Ruff (0.15.7)

build-scripts/fix_kernelsu_49_v39.py

[warning] 23-23: Unnecessary mode argument

Remove mode argument

(UP015)

---

[warning] 54-54: Unnecessary else after return statement

Remove unnecessary else

(RET505)

---

[warning] 61-61: Unnecessary mode argument

Remove mode argument

(UP015)

---

[warning] 148-148: Loop control variable declared not used within loop body

Rename unused declared to _declared

(B007)

---

[warning] 153-153: Unnecessary else after return statement

Remove unnecessary else

(RET505)

🔇 Additional comments (2)

build-scripts/fix_kernelsu_49_v39.py (2)

17-55: LGTM! The regex pattern correctly fixes the lookbehind issue from v38.

The pattern (?<!sys_)(?<!try_)\bumount properly excludes already-prefixed calls, unlike v38's malformed (?<!)try_). The capture group and replacement logic are correct.

Minor style notes from static analysis (optional to address):

• Line 23: open(file_path, 'r') → open(file_path) (mode 'r' is default)
• Line 54: else after return is unnecessary

---

58-86: LGTM! Declaration insertion logic is sound for kernel source files.

The check for existing declarations and insertion after the last #include is appropriate. Minor note: if no #include lines exist, insert_idx would be 0 (inserting at file start), but this is an unlikely edge case for kernel C files.