truncate: eliminate duplicate stat() syscall #9527

Jean-Christian-Cirstea · 2025-11-30T01:08:38Z

When no reference file is given, truncate performs two stat() syscalls:

The first one to retrieve the size of the file.
The second one to check if the file if a pipe.

This commit fixes this issue by restructuring the code. The main change resides in the elimination of the initial match performed at the top level to select between different scenarios. Instead, the new implementation:

Check if a reference file has been given. If so, retrieve its size.
Determine the given mode. If no mode has been given, TruncateMode::Extend(0) is implied.
Retrieve the size of the file, while checking if it is a pipe. This is the key point of this PR.
Determine the truncate size based on the reference size and the mode. If no reference file has been provided, use the size of the file to be resized as a reference size.
Finally, perform ftruncate() to resize the file.

Jean-Christian-Cirstea · 2025-11-30T01:09:47Z

src/uu/truncate/src/truncate.rs

+    /// assert_eq!(mode.to_size(fsize), None);
    /// ```
-    fn to_size(&self, fsize: u64) -> u64 {
+    fn to_size(&self, fsize: u64) -> Option<u64> {


Previously, this method could panic if rounding to 0. Now, the method returns None if rounding to 0.

would it be possible to add a unit test for this? thanks

Already added. See lines 359:361.

Jean-Christian-Cirstea · 2025-11-30T01:11:38Z

src/uu/truncate/src/truncate.rs

+            Self::RoundDown(size) => fsize.checked_rem(*size).map(|remainder| fsize - remainder),
+            Self::RoundUp(size) => fsize.checked_rem(*size).map(|remainder| fsize + remainder),


As previously stated, calling this method when rounding to 0 resulted in a panic. Following the guideline of avoiding panicking code, use checked_rem() method.

Jean-Christian-Cirstea · 2025-11-30T01:12:18Z

src/uu/truncate/src/truncate.rs

+    fn is_absolute(&self) -> bool {
+        matches!(self, Self::Absolute(_))
+    }


Tiny helper function to determine if self is absolute. Useful when checking if an absolute mode was provided with a reference file.

Jean-Christian-Cirstea · 2025-11-30T01:13:00Z

src/uu/truncate/src/truncate.rs

-    #[cfg(unix)]
-    if let Ok(metadata) = metadata(path) {
-        if metadata.file_type().is_fifo() {
-            return Err(USimpleError::new(
-                1,
-                translate!("truncate-error-cannot-open-no-device", "filename" => filename.to_string_lossy().quote()),
-            ));
-        }
-    }


This check was moved to line 212.

Jean-Christian-Cirstea · 2025-11-30T01:14:00Z

src/uu/truncate/src/truncate.rs

+    let file_size = match metadata(path) {
+        Ok(metadata) => {
+            #[cfg(unix)]
+            if metadata.file_type().is_fifo() {
+                return Err(USimpleError::new(
+                    1,
+                    translate!("truncate-error-cannot-open-no-device", "filename" => filename.to_string_lossy().quote()),
+                ));
+            }
+            metadata.len()
        }
-        Ok(m) => m,
+        Err(_) => 0,
    };


Retrieve the size of the file, while checking if it is a pipe. One single stat() syscall required.

maybe add this as a comment in the code?

Added comments.

Jean-Christian-Cirstea · 2025-11-30T01:15:30Z

src/uu/truncate/src/truncate.rs

+    // Omitting the mode is equivalent to extending a file by 0 bytes.
+    let mode = match size_string {
+        Some(string) => match parse_mode_and_size(string) {
+            Err(error) => {
+                return Err(USimpleError::new(
+                    1,
+                    translate!("truncate-error-invalid-number", "error" => error),
+                ));
+            }
+            Ok(mode) => mode,
+        },
+        None => TruncateMode::Extend(0),
+    };


Specifying no mode implies TruncateMode::Extend(0).

same, better in the code as comment :)

There is a comment. Look carefully at line 263. Est-ce que j'aurais dû l'écrire en français pour qu'il soit plus clair ? :)

Oups pardon :)

Jean-Christian-Cirstea · 2025-11-30T01:17:46Z

src/uu/truncate/src/truncate.rs

-            Self::Reduce(size) => {
-                if *size > fsize {
-                    0
-                } else {
-                    fsize - size
-                }
-            }


The standard library provides saturating_sub(). No need to implement its logic by ourselves.

github-actions · 2025-11-30T08:29:56Z

GNU testsuite comparison:

Skip an intermittent issue tests/tail/overlay-headers (fails in this run but passes in the 'main' branch)

Signed-off-by: Jean-Christian CÎRSTEA <[email protected]>

1. The documentation for parse_mode_and_size() has been updated. 2. Added documentation for TruncateMode::is_absolute(). Signed-off-by: Jean-Christian CÎRSTEA <[email protected]>

github-actions · 2025-11-30T10:07:50Z

GNU testsuite comparison:

Skip an intermittent issue tests/tail/overlay-headers (fails in this run but passes in the 'main' branch)
Skipping an intermittent issue tests/misc/tee (passes in this run but fails in the 'main' branch)

Jean-Christian-Cirstea commented Nov 30, 2025

View reviewed changes

Jean-Christian-Cirstea added 2 commits November 30, 2025 11:49

truncate: Avoid duplicate stat() syscall

e5e64f9

Signed-off-by: Jean-Christian CÎRSTEA <[email protected]>

truncate: updated documentation

abcae00

1. The documentation for parse_mode_and_size() has been updated. 2. Added documentation for TruncateMode::is_absolute(). Signed-off-by: Jean-Christian CÎRSTEA <[email protected]>

Jean-Christian-Cirstea force-pushed the truncate branch from a5efbf3 to abcae00 Compare November 30, 2025 09:50

Jean-Christian-Cirstea requested a review from sylvestre November 30, 2025 10:41

		Self::RoundDown(size) => fsize.checked_rem(*size).map(\|remainder\| fsize - remainder),
		Self::RoundUp(size) => fsize.checked_rem(*size).map(\|remainder\| fsize + remainder),

Uh oh!

truncate: eliminate duplicate stat() syscall #9527

Are you sure you want to change the base?

truncate: eliminate duplicate stat() syscall #9527

Uh oh!

Conversation

Jean-Christian-Cirstea commented Nov 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Jean-Christian-Cirstea Nov 30, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

github-actions bot commented Nov 30, 2025

Uh oh!

github-actions bot commented Nov 30, 2025

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Jean-Christian-Cirstea commented Nov 30, 2025 •

edited

Loading

Jean-Christian-Cirstea Nov 30, 2025 •

edited

Loading