From a9391c90f07a564693bb1994f4f8fe6470efb3a2 Mon Sep 17 00:00:00 2001
From: Craig Jennings <c@cjennings.net>
Date: Fri, 24 Jul 2026 23:16:56 -0500
Subject: fix(installer): harden the lock path against the AMD-iGPU DPMS
 lockout

An idle lock on this Strix Halo box wedged the whole session: hyprlock died and the compositor stayed locked with no prompt, recoverable only from a console. It's a documented AMD-integrated-Radeon failure (hyprlock#953, Hyprland#5822) -- a display power cycle via DPMS invalidates the GPU resources the lock client holds, so hyprlock loses its surface and exits without unlocking. No coredump, no OOM; the GPU pulls the rug out.

Two installer changes, both scoped and tested:

update_grub_cmdline adds amdgpu.runpm=0 on AMD machines only. Disabling GPU runtime power management keeps those resources valid across a display cycle -- the root fix. A no-op on Intel/NVIDIA, and it rides the existing merge so no boot-critical token is touched.

configure_hyprlock_pam writes a complete PAM stack. The hyprlock package ships only `auth include login`, leaving account and session uninitialised so pam_end() crashes on cleanup -- a separate documented lockout cause. All three phases now resolve through login, inheriting the keyring the graphical login uses.

CALL_SITES pins both new wirings. 372 unit tests, exit 0; each addition proven by reverting it.
---
 tests/installer-steps/test_grub_cmdline.py | 42 ++++++++++++++++++++++++++++++
 1 file changed, 42 insertions(+)

(limited to 'tests/installer-steps/test_grub_cmdline.py')

diff --git a/tests/installer-steps/test_grub_cmdline.py b/tests/installer-steps/test_grub_cmdline.py
index 8c0b5b0..4c2d202 100644
--- a/tests/installer-steps/test_grub_cmdline.py
+++ b/tests/installer-steps/test_grub_cmdline.py
@@ -67,6 +67,48 @@ class MergeGrubCmdline(unittest.TestCase):
         self.assertIn("zfs=zroot/ROOT/default", out)
 
 
+def run_with_gpu(body, gpu):
+    # detect_gpu_vendors is stubbed to a chosen vendor list so the AMD-only
+    # amdgpu.runpm=0 addition can be exercised without real /sys reads.
+    stub = 'detect_gpu_vendors() { printf "%s\\n" %s; }\n' % ("%s", "'" + gpu + "'")
+    stub = 'detect_gpu_vendors() { printf "%s" "$GPU"; [ -n "$GPU" ] && echo; }\n'
+    script = ('logfile=/dev/null\nerror_warn() { echo "WARN: $1"; return 1; }\n'
+              + stub + EXTRACT + "\n" + body + "\n")
+    return subprocess.run(["bash", "-c", script], capture_output=True, text=True,
+                          timeout=10, env={**os.environ, "GPU": gpu})
+
+
+class AmdRunpmCmdline(unittest.TestCase):
+    """amdgpu.runpm=0 is added on AMD machines only: the AMD iGPU invalidates
+    hyprlock's GPU resources on a display power cycle, wedging the lock
+    (hyprlock#953). Disabling GPU runtime PM keeps them valid. A no-op on Intel
+    or NVIDIA."""
+
+    def _file(self, gpu):
+        import tempfile
+        fd, path = tempfile.mkstemp(prefix="grub-runpm-")
+        os.close(fd)
+        with open(path, "w") as f:
+            f.write('GRUB_CMDLINE_LINUX_DEFAULT="rw quiet"\n')
+        self.addCleanup(os.remove, path)
+        run_with_gpu(f'update_grub_cmdline "{path}"', gpu)
+        with open(path) as f:
+            return f.read()
+
+    def test_amd_gets_runpm_disabled(self):
+        self.assertIn("amdgpu.runpm=0", self._file("amd"))
+
+    def test_intel_does_not(self):
+        self.assertNotIn("amdgpu.runpm", self._file("intel"))
+
+    def test_no_gpu_detected_does_not(self):
+        self.assertNotIn("amdgpu.runpm", self._file(""))
+
+    def test_amd_with_nvidia_still_gets_it(self):
+        # A hybrid box with an AMD part present still wants the fix.
+        self.assertIn("amdgpu.runpm=0", self._file("amd\nnvidia"))
+
+
 class UpdateGrubCmdline(unittest.TestCase):
     def run_update(self, grub_body):
         with tempfile.TemporaryDirectory() as d:
-- 
cgit v1.2.3