Proof Portal

Project overview

Omarchy

ProbeLabs65 findings · 95 requirements

A proof layer — requirements, tests and verified fixes — for two of Omarchy's subsystems: the application menu (launcher scripts, QML model, JSONC config, search and selection) and the lock screen (lock scripts, QML, PAM authentication). Scope is deliberately limited to those components of omacom/omarchy; the rest of the distribution is not covered.

Inbox
ReportR-4823GitHub issuePublicReceived

omarchy-windows-vm: setgid on ~/Windows silently blocks every privileged VM action

Reported by ACooperDev · filed Sep 1, 2026Closed upstream on 2026-09-20 by bjarneo via an automated message, as a duplicate of #9334 (R-1173). GitHub close reason: completed.

Received and recorded verbatim. Nothing has been checked yet.

At a glance

1 · What is claimed?
Only the report text below — no validator has restated it as checkable claims yet.
2 · What does it affect?
Not yet mapped to any requirement or component.
3 · Is it reproducible?
No validation has been run yet.
4 · What remains unclear?
  • Not validated yet: no claims, no reproducer.
5 · What happens next?
Our staff decide whether the report is valid. Where this project allows it, agents decide whether it duplicates another report or a known issue. No decision has been recorded yet.

Original report

omarchy-windows-vm: setgid on ~/Windows silently blocks every privileged VM actionrevision 1 · captured 2026-10-04 12:03 UTC via poll

What happened

Every privileged omarchy windows vm operation (remove, launch, stop) fails
when either VM source directory — ~/.windows or ~/Windows — has the setgid bit
set. The failure is silent: the privileged helper returns 1 without printing a
reason, so the user only sees the generic wrapper message.

$ omarchy windows vm remove
Removing Windows VM...
❌ Windows VM removal stopped before user-side cleanup; inspect the VM data before retrying.

~/Windows was mode 2700 (drwx--S---). No other diagnostic is printed, so
there is nothing pointing at the mode as the cause.

Root cause

omarchy-windows-vm, prepare_caller_mounts():

chmod 0700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {
  ...
}
storage_mode=$(stat -Lc '%a' "/proc/$BASHPID/fd/$storage_fd" 2>/dev/null) || storage_mode=""
shared_mode=$(stat -Lc '%a' "/proc/$BASHPID/fd/$shared_fd" 2>/dev/null) || shared_mode=""
if [[ $storage_mode != 700 || $shared_mode != 700 ]]; then
  exec {storage_fd}<&-
  exec {shared_fd}<&-
  return 1
fi

GNU chmod does not clear setuid/setgid on a directory for a numeric mode of
four digits or fewer — 0700 leaves the setgid bit untouched. stat -Lc '%a'
then reports 2700, the != 700 comparison fails, and the function returns 1.

The hardening step can therefore never satisfy its own postcondition, and the
mode check is unreachable-by-success for any setgid source directory.

This blocks every privileged path, since they all funnel through
assert_mounts_safe → prepare_caller_mounts:

  • __priv_remove() — assert_mounts_safe || return 1
  • __priv_up() — assert_mounts_safe && dc up -d
  • __priv_up_wait() — assert_mounts_safe || return 1

Nothing is destroyed (the check runs before dc down and the find -delete), but
the VM becomes permanently un-launchable and un-removable through the CLI.

Minimal reproduction

Independent of the VM — this is just the predicate:

d=$(mktemp -d); chmod 2700 "$d"
exec {fd}<"$d/."
chmod 0700 -- "/proc/$BASHPID/fd/$fd"
stat -Lc '%a' "/proc/$BASHPID/fd/$fd"   # prints 2700, expected 700
exec {fd}<&-

Observed on coreutils 9.11-2:

chmod formresulting mode
07002700 ❌
u=rwx,go=2700 ❌
00700700 ✅
g-s700 ✅

Note that the symbolic form u=rwx,go= does not clear it either — GNU chmod
preserves setuid/setgid on directories unless they are named explicitly.

End-to-end:

  1. chmod g+s ~/Windows (or have a pre-existing setgid ~/Windows before install —

install_windows only creates the directory when it is absent:
[[ -e $shared || -L $shared ]] || install -d -m 0700 -- "$shared", so an
existing setgid directory is adopted as-is)
2. omarchy windows vm install
3. omarchy windows vm remove → fails with the generic message above
4. omarchy windows vm launch → fails the same way

Expected

Either the hardening actually clears the special bits, or the failure explains
itself. Both would be better.

Suggested fix

-  chmod 0700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {
+  # Numeric modes of four digits or fewer leave setuid/setgid on a directory
+  # alone; the fifth digit is what clears them, and the mode check below
+  # requires exactly 700.
+  chmod 00700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {

and give the mode check a diagnostic so it is not silent:

   if [[ $storage_mode != 700 || $shared_mode != 700 ]]; then
+    echo "omarchy-windows-vm: VM source directories must be mode 700 (storage=$storage_mode shared=$shared_mode)" >&2
     exec {storage_fd}<&-
     exec {shared_fd}<&-
     return 1
   fi

The same silent-return 1 pattern appears elsewhere in prepare_caller_mounts
and in __priv_remove (the EXPECTED_STORAGE/EXPECTED_SHARED equality checks),
which is why the wrapper's generic message is all the user ever gets. Worth a
pass, though the setgid case is the one that actually bites.

Workaround

chmod g-s ~/Windows ~/.windows

System details

  • Omarchy 4.0.2-1
  • Kernel 7.1.9-arch1-2
  • coreutils 9.11-2
  • Docker 29.7.2
  • Hyprland 0.56.2
  • CPU: Intel Core i9-10885H
  • GPU: Intel CometLake-H GT2 [UHD Graphics] / NVIDIA TU104GLM [Quadro RTX 5000 Mobile / Max-Q]
  • Sudoless Docker: not enabled (privileged actions go through pkexec)
View raw text
### What happened

Every privileged `omarchy windows vm` operation (`remove`, `launch`, `stop`) fails
when either VM source directory — `~/.windows` or `~/Windows` — has the setgid bit
set. The failure is silent: the privileged helper returns 1 without printing a
reason, so the user only sees the generic wrapper message.

```
$ omarchy windows vm remove
Removing Windows VM...
❌ Windows VM removal stopped before user-side cleanup; inspect the VM data before retrying.
```

`~/Windows` was mode `2700` (`drwx--S---`). No other diagnostic is printed, so
there is nothing pointing at the mode as the cause.

### Root cause

`omarchy-windows-vm`, `prepare_caller_mounts()`:

```bash
chmod 0700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {
  ...
}
storage_mode=$(stat -Lc '%a' "/proc/$BASHPID/fd/$storage_fd" 2>/dev/null) || storage_mode=""
shared_mode=$(stat -Lc '%a' "/proc/$BASHPID/fd/$shared_fd" 2>/dev/null) || shared_mode=""
if [[ $storage_mode != 700 || $shared_mode != 700 ]]; then
  exec {storage_fd}<&-
  exec {shared_fd}<&-
  return 1
fi
```

GNU `chmod` does not clear setuid/setgid on a **directory** for a numeric mode of
four digits or fewer — `0700` leaves the setgid bit untouched. `stat -Lc '%a'`
then reports `2700`, the `!= 700` comparison fails, and the function returns 1.

The hardening step can therefore never satisfy its own postcondition, and the
mode check is unreachable-by-success for any setgid source directory.

This blocks every privileged path, since they all funnel through
`assert_mounts_safe` → `prepare_caller_mounts`:

- `__priv_remove()` — `assert_mounts_safe || return 1`
- `__priv_up()` — `assert_mounts_safe && dc up -d`
- `__priv_up_wait()` — `assert_mounts_safe || return 1`

Nothing is destroyed (the check runs before `dc down` and the `find -delete`), but
the VM becomes permanently un-launchable and un-removable through the CLI.

### Minimal reproduction

Independent of the VM — this is just the predicate:

```bash
d=$(mktemp -d); chmod 2700 "$d"
exec {fd}<"$d/."
chmod 0700 -- "/proc/$BASHPID/fd/$fd"
stat -Lc '%a' "/proc/$BASHPID/fd/$fd"   # prints 2700, expected 700
exec {fd}<&-
```

Observed on coreutils 9.11-2:

| chmod form  | resulting mode |
|-------------|----------------|
| `0700`      | `2700` ❌ |
| `u=rwx,go=` | `2700` ❌ |
| `00700`     | `700` ✅ |
| `g-s`       | `700` ✅ |

Note that the symbolic form `u=rwx,go=` does not clear it either — GNU chmod
preserves setuid/setgid on directories unless they are named explicitly.

End-to-end:

1. `chmod g+s ~/Windows` (or have a pre-existing setgid `~/Windows` before install —
   `install_windows` only creates the directory when it is absent:
   `[[ -e $shared || -L $shared ]] || install -d -m 0700 -- "$shared"`, so an
   existing setgid directory is adopted as-is)
2. `omarchy windows vm install`
3. `omarchy windows vm remove` → fails with the generic message above
4. `omarchy windows vm launch` → fails the same way

### Expected

Either the hardening actually clears the special bits, or the failure explains
itself. Both would be better.

### Suggested fix

```diff
-  chmod 0700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {
+  # Numeric modes of four digits or fewer leave setuid/setgid on a directory
+  # alone; the fifth digit is what clears them, and the mode check below
+  # requires exactly 700.
+  chmod 00700 -- "/proc/$BASHPID/fd/$storage_fd" "/proc/$BASHPID/fd/$shared_fd" || {
```

and give the mode check a diagnostic so it is not silent:

```diff
   if [[ $storage_mode != 700 || $shared_mode != 700 ]]; then
+    echo "omarchy-windows-vm: VM source directories must be mode 700 (storage=$storage_mode shared=$shared_mode)" >&2
     exec {storage_fd}<&-
     exec {shared_fd}<&-
     return 1
   fi
```

The same silent-`return 1` pattern appears elsewhere in `prepare_caller_mounts`
and in `__priv_remove` (the `EXPECTED_STORAGE`/`EXPECTED_SHARED` equality checks),
which is why the wrapper's generic message is all the user ever gets. Worth a
pass, though the setgid case is the one that actually bites.

### Workaround

```bash
chmod g-s ~/Windows ~/.windows
```

### System details

- Omarchy 4.0.2-1
- Kernel 7.1.9-arch1-2
- coreutils 9.11-2
- Docker 29.7.2
- Hyprland 0.56.2
- CPU: Intel Core i9-10885H
- GPU: Intel CometLake-H GT2 [UHD Graphics] / NVIDIA TU104GLM [Quadro RTX 5000 Mobile / Max-Q]
- Sudoless Docker: not enabled (privileged actions go through pkexec)

Validation

Not checked yet. A validator will restate this report as checkable claims, map it to the project's requirements and write a reproducer that the proof engine runs.

Linked

Not linked to any known issue, other report or change yet.

Decision

Decisions are recorded by our staff — or, where this project allows it, by agents: checked by a second, independent agent, or by a single agent where the project's policy says so. Any decision reached so far is shown below.

No decision has been recorded yet.