fix(sort): GNU sort-continue.sh test

src/uu/sort/src/merge.rs

@@ -30,7 +30,7 @@ use uucore::error::{FromIo, UResult};
 use crate::{
     GlobalSettings, Output, SortError,
     chunks::{self, Chunk, RecycledChunk},
-    compare_by, open,
+    compare_by, fd_soft_limit, open,
     tmp_dir::TmpDirWrapper,
 };
 
@@ -62,6 +62,26 @@ fn replace_output_file_in_input_files(
     Ok(())
 }
 
+fn effective_merge_batch_size(settings: &GlobalSettings) -> usize {
+    const MIN_BATCH_SIZE: usize = 2;
+    const RESERVED_STDIO: usize = 3;
+    const RESERVED_OUTPUT: usize = 1;
+    const SAFETY_MARGIN: usize = 1;
+    let mut batch_size = settings.merge_batch_size.max(MIN_BATCH_SIZE);
+
+    if let Some(limit) = fd_soft_limit() {
+        let reserved = RESERVED_STDIO + RESERVED_OUTPUT + SAFETY_MARGIN;
+        let available_inputs = limit.saturating_sub(reserved);
+        if available_inputs >= MIN_BATCH_SIZE {
+            batch_size = batch_size.min(available_inputs);
+        } else {
+            batch_size = MIN_BATCH_SIZE;
+        }
+    }
+
+    batch_size
+}
+
 /// Merge pre-sorted `Box<dyn Read>`s.
 ///
 /// If `settings.merge_batch_size` is greater than the length of `files`, intermediate files will be used.
@@ -94,18 +114,21 @@ pub fn merge_with_file_limit<
     output: Output,
     tmp_dir: &mut TmpDirWrapper,
 ) -> UResult<()> {
-    if files.len() <= settings.merge_batch_size {
+    let batch_size = effective_merge_batch_size(settings);
+    debug_assert!(batch_size >= 2);
+
+    if files.len() <= batch_size {
         let merger = merge_without_limit(files, settings);
         merger?.write_all(settings, output)
     } else {
         let mut temporary_files = vec![];
-        let mut batch = vec![];
+        let mut batch = Vec::with_capacity(batch_size);
         for file in files {
             batch.push(file);
-            if batch.len() >= settings.merge_batch_size {
-                assert_eq!(batch.len(), settings.merge_batch_size);
+            if batch.len() >= batch_size {
+                assert_eq!(batch.len(), batch_size);
                 let merger = merge_without_limit(batch.into_iter(), settings)?;
-                batch = vec![];
+                batch = Vec::with_capacity(batch_size);
 
                 let mut tmp_file =
                     Tmp::create(tmp_dir.next_file()?, settings.compress_prog.as_deref())?;
@@ -115,7 +138,7 @@ pub fn merge_with_file_limit<
         }
         // Merge any remaining files that didn't get merged in a full batch above.
         if !batch.is_empty() {
-            assert!(batch.len() < settings.merge_batch_size);
+            assert!(batch.len() < batch_size);
             let merger = merge_without_limit(batch.into_iter(), settings)?;
 
             let mut tmp_file =

src/uu/sort/src/sort.rs

@@ -26,7 +26,7 @@ use custom_str_cmp::custom_str_cmp;
 use ext_sort::ext_sort;
 use fnv::FnvHasher;
 #[cfg(target_os = "linux")]
-use nix::libc::{RLIMIT_NOFILE, getrlimit, rlimit};
+use nix::libc;
 use numeric_str_cmp::{NumInfo, NumInfoParseSettings, human_numeric_str_cmp, numeric_str_cmp};
 use rand::{Rng, rng};
 use rayon::prelude::*;
@@ -1073,17 +1073,30 @@ fn make_sort_mode_arg(mode: &'static str, short: char, help: String) -> Arg {
 }
 
 #[cfg(target_os = "linux")]
-fn get_rlimit() -> UResult<usize> {
-    let mut limit = rlimit {
+pub(crate) fn fd_soft_limit() -> Option<usize> {
+    let mut limit = libc::rlimit {
         rlim_cur: 0,
         rlim_max: 0,
     };
-    match unsafe { getrlimit(RLIMIT_NOFILE, &raw mut limit) } {
-        0 => Ok(limit.rlim_cur as usize),
-        _ => Err(UUsageError::new(2, translate!("sort-failed-fetch-rlimit"))),
+
+    let result = unsafe { libc::getrlimit(libc::RLIMIT_NOFILE, &raw mut limit) };
+    if result == 0 {
+        let current = limit.rlim_cur;
+        if current == libc::RLIM_INFINITY {
+            None
+        } else {
+            usize::try_from(current).ok()
+        }
+    } else {
+        None
     }
 }
 
+#[cfg(not(target_os = "linux"))]
+pub(crate) fn fd_soft_limit() -> Option<usize> {
+    None
+}
+
 const STDIN_FILE: &str = "-";
 #[cfg(target_os = "linux")]
 const LINUX_BATCH_DIVISOR: usize = 4;
@@ -1096,12 +1109,12 @@ fn default_merge_batch_size() -> usize {
     #[cfg(target_os = "linux")]
     {
         // Adjust merge batch size dynamically based on available file descriptors.
-        match get_rlimit() {
-            Ok(limit) => {
+        match fd_soft_limit() {
+            Some(limit) => {
                 let usable_limit = limit.saturating_div(LINUX_BATCH_DIVISOR);
                 usable_limit.clamp(LINUX_BATCH_MIN, LINUX_BATCH_MAX)
             }
-            Err(_) => 64,
+            None => 64,
         }
     }
 
@@ -1279,7 +1292,15 @@ pub fn uumain(args: impl uucore::Args) -> UResult<()> {
 
                         translate!(
                             "sort-maximum-batch-size-rlimit",
-                            "rlimit" =>  get_rlimit()?
+                            "rlimit" => {
+                                let Some(rlimit) = fd_soft_limit() else {
+                                    return Err(UUsageError::new(
+                                        2,
+                                        translate!("sort-failed-fetch-rlimit"),
+                                    ));
+                                };
+                                rlimit
+                            }
                         )
                     }
                     #[cfg(not(target_os = "linux"))]