df: report an over-mounted device instead of falling back to statfs by montezuma-p · Pull Request #15977 · uutils/coreutils

montezuma-p · 2026-10-29T15:34:22Z

df DEVICE printed the row of the file system holding the device node (usually /dev) and exited 0 when the device's mount point was over-mounted, or when that mount point could not be statfs'd. The statfs fallback added for a mount table with no entry covering the path ran on every error of the lookup, including FsError::OverMounted.

This moves the decision into Filesystem::from_path(), the only place that can tell "no entry covers the path" from an error about the mount that was found, and lets the latter reach the caller as it did in 0.10.0. Three unit tests with a synthetic mount table cover the over-mounted case, the unreachable mount point, and the fallback still working when nothing covers the path.

Fixes #15799

Copilot

🟢 Approval recommended

No active review issues were identified for the change restricting df’s statfs fallback to operands without a matching mount-table entry.

No open issues 🎉

What changed in this PR

Fixes #15799 by restricting df’s statfs fallback to cases where no mount-table entry covers the operand, preserving errors for over-mounted devices and inaccessible mount points.

Changes:

Moves the fallback decision into Filesystem::from_path().
Removes the Unix caller’s catch-all fallback.
Adds three unit tests covering over-mount errors, mount-point statfs failures, and successful fallback.

File	Description
`src/uu/df/src/platform/unix.rs`	Removes catch-all fallback, exposes the fallback helper, and adds regression tests.
`src/uu/df/src/filesystem.rs`	Limits fallback to missing mount entries and documents error propagation.

_{Review effort: Balanced}

github-actions · 2026-10-29T15:48:23Z

GNU testsuite comparison:

Congrats! The gnu test tests/df/over-mount-device is no longer failing!

sylvestre · 2026-11-02T07:53:52Z

+
+        assert_eq!(
+            filesystem_for_path(&mounts, false, "/dev/null").unwrap_err(),
+            FsError::MountMissing


exit 1 now, good, but the user gets "no file systems processed" here while GNU says "Permission denied" on the mount point
maybe keep the io error for a follow-up?

Done in this PR (8c44d93): when statfs fails on the mount point, df now reports it like GNU, e.g. df: /mnt: Permission denied, and still exits 1.

`df DEVICE` printed the row of the file system holding the device node (usually `/dev`) and exited 0 when the device's mount point was over-mounted, or when its mount point could not be statfs'd. The statfs fallback in `filesystem_for_path()` ran on every error of the mount table lookup, not only when no entry covered the operand. Move the decision into `Filesystem::from_path()`, the one place that can tell "no entry covers the path" from an error about the mount that was found, and let the latter reach the caller as before. Fixes uutils#15799

Like GNU, print the statfs error for the mount point, e.g. "Permission denied", instead of "no file systems processed".

Copilot

Warning

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Copilot review overview

4 open issues

Filesystem::from_path() now unconditionally references platform::from_path_direct_with_mounts.… · New
Storing the underlying statfs failure as a String loses structured error information… · New
This doc comment describes Unix-only behavior (statfs fallback and FsError::OverMounted). On… · New
This test compares an OsString (path) to a string literal, relying on platform-specific… · New

_{Review effort: Lite}

montezuma-p · 2026-11-03T23:21:40Z

+        match mount_info_from_path(mounts, &path, canonicalize) {
+            Ok(mount_info) => platform::filesystem_from_mount(mounts, mount_info, Some(file)),
+            // Only an empty lookup may fall back to `statfs`: an error about
+            // the mount that was found is the answer for `path`.
+            Err(FsError::MountMissing) => platform::from_path_direct_with_mounts(mounts, path),
+            Err(e) => Err(e),
+        }


Filesystem::from_path() is #[cfg(unix)], as is FsError::StatFailed, so Windows never compiles this arm or references from_path_direct_with_mounts.

+    /// `statfs` failed on the mount point at `path` with `error`.
+    #[cfg(not(windows))]
+    StatFailed {
+        path: OsString,
+        error: String,
+    },


+    /// longest mount directory is returned. If no element of `mounts`
+    /// covers `path`, the filesystem is looked up through `statfs`
+    /// instead. An error about the mount that was found, such as
+    /// [`FsError::OverMounted`], is returned as is.


+
+        assert!(matches!(
+            filesystem_for_path(&mounts, false, "/dev/null").unwrap_err(),
+            FsError::StatFailed { path, .. } if path == "/non-existent-dir"


Copilot

Warning

Copilot couldn't run its full agentic review because it didn't start before the timeout. Make sure your repository has a runner available, or add a copilot-code-review.yml file specifying one with the runs-on attribute. See the docs for more details.

Copilot review overview

6 open issues

MountInfo::dev_name is a String (see MountInfo usage elsewhere), so… · New
Filesystem::from_path() now unconditionally references platform::from_path_direct_with_mounts.…
path in FsError::StatFailed is an OsString, and comparing it directly to a &str (`path ==… · New
This doc comment describes Unix-only behavior (statfs fallback and FsError::OverMounted). On…
Storing the underlying statfs failure as a String loses structured error information…
This test compares an OsString (path) to a string literal, relying on platform-specific…

_{Review effort: Lite}

+/// The path `statfs` is called on to query the filesystem at `mount_info`.
+fn stat_path(mount_info: &MountInfo) -> &OsStr {
+    if mount_info.mount_dir.is_empty() {
        mount_info.dev_name.as_ref()
    } else {
        mount_info.mount_dir.as_os_str()
-    };
-    Some(FsUsage::new(statfs(stat_path).ok()?))
+    }
 }


+
+        assert!(matches!(
+            filesystem_for_path(&mounts, false, "/dev/null").unwrap_err(),
+            FsError::StatFailed { path, .. } if path == "/non-existent-dir"


sylvestre · 2026-11-04T11:27:04Z

Thanks for your PR

Copilot AI balanced review requested due to automatic review settings October 29, 2026 15:34

Copilot started reviewing on behalf of montezuma-p October 29, 2026 15:34

View session

Copilot AI reviewed Oct 29, 2026

View reviewed changes

sylvestre reviewed Nov 2, 2026

View reviewed changes

sylvestre mentioned this pull request Nov 3, 2026

GNU coreutils 9.12: remaining test failures - Nov 2026 #14705

Open

32 of 76 tasks

montezuma-p requested review from sylvestre and a balanced review from Copilot November 3, 2026 22:59

montezuma-p added 2 commits November 4, 2026 00:05

df: report why the mount point could not be statfs'd

47b7daf

Like GNU, print the statfs error for the mount point, e.g. "Permission denied", instead of "no file systems processed".

sylvestre force-pushed the df-overmounted-error branch from 8c44d93 to 47b7daf Compare November 3, 2026 23:05

Copilot AI reviewed Nov 3, 2026

View reviewed changes

Copilot AI balanced review requested due to automatic review settings November 3, 2026 23:17

Copilot AI reviewed Nov 3, 2026

View reviewed changes

Copilot stopped reviewing on behalf of montezuma-p due to an error November 4, 2026 00:43
The job was not acquired by a runner even after multiple attempts.

Copilot stopped reviewing on behalf of montezuma-p due to an error November 4, 2026 01:15
The job was not acquired by a runner even after multiple attempts.

sylvestre merged commit 56cb9ea into uutils:main Nov 4, 2026
99 checks passed

Navigation Menu

Repository navigation

df: report an over-mounted device instead of falling back to statfs - #15977

df: report an over-mounted device instead of falling back to statfs#15977
sylvestre merged 2 commits into
uutils:mainfrom
montezuma-p:df-overmounted-error

montezuma-p commented Oct 29, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

github-actions Bot commented Oct 29, 2026

Uh oh!

sylvestre Nov 2, 2026

Uh oh!

montezuma-p Nov 3, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

montezuma-p Nov 3, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

sylvestre commented Nov 4, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

montezuma-p commented Oct 29, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

🟢 Approval recommended

Uh oh!

github-actions Bot commented Oct 29, 2026

Uh oh!

sylvestre Nov 2, 2026

Choose a reason for hiding this comment

Uh oh!

montezuma-p Nov 3, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

Uh oh!

montezuma-p Nov 3, 2026

Choose a reason for hiding this comment

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Copilot review overview

Uh oh!

Uh oh!

sylvestre commented Nov 4, 2026

Uh oh!

Reviewers

Uh oh!

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants