Merge pull request #482 from rustcoreutils/updates

HashMap and Vec micro-opts

Author: jgarzik

awk/interpreter/mod.rs

@@ -61,7 +61,7 @@ fn strtod(s: &str) -> f64 {
 }
 
 fn gather_values(stack: &mut Stack, count: u16) -> Result<Vec<AwkValue>, String> {
-    let mut values = Vec::new();
+    let mut values = Vec::with_capacity(count as usize);
     for _ in 0..count {
         values.push(stack.pop_scalar_value()?);
     }
@@ -73,7 +73,7 @@ fn print_to_string(
     argc: u16,
     global_env: &GlobalEnv,
 ) -> Result<AwkString, String> {
-    let mut values = Vec::new();
+    let mut values = Vec::with_capacity(argc as usize);
     for _ in 0..argc {
         values.push(
             stack

cc/arch/x86_64/codegen.rs

@@ -1898,7 +1898,7 @@ impl X86_64CodeGen {
         // First pass: determine which args go on stack
         // For variadic calls: variadic INTEGER args go on stack, variadic FLOAT args use XMM
         // For non-variadic calls: overflow args go on stack
-        let mut stack_arg_indices = Vec::new();
+        let mut stack_arg_indices = Vec::with_capacity(insn.src.len());
         let mut temp_int_idx = 0;
         let mut temp_fp_idx = 0;
 
@@ -3734,22 +3734,23 @@ impl X86_64CodeGen {
         // Build operand strings for asm substitution
         // For constraints requiring specific registers (a,b,c,d,S,D), we use those registers
         // and emit mov instructions to/from the actual locations.
-        let mut operand_regs: Vec<Option<Reg>> = Vec::new();
-        let mut operand_mem: Vec<Option<String>> = Vec::new();
-        let mut operand_names: Vec<Option<String>> = Vec::new();
+        let operand_count = asm_data.outputs.len() + asm_data.inputs.len();
+        let mut operand_regs: Vec<Option<Reg>> = Vec::with_capacity(operand_count);
+        let mut operand_mem: Vec<Option<String>> = Vec::with_capacity(operand_count);
+        let mut operand_names: Vec<Option<String>> = Vec::with_capacity(operand_count);
 
         // Track which outputs need to be moved from specific registers after asm
         // (output_idx, specific_reg, actual_loc)
-        let mut output_moves: Vec<(usize, Reg, Loc)> = Vec::new();
+        let mut output_moves: Vec<(usize, Reg, Loc)> = Vec::with_capacity(asm_data.outputs.len());
 
         // Track which inputs need to be moved to specific registers before asm
         // (specific_reg, actual_loc)
-        let mut input_moves: Vec<(Reg, Loc)> = Vec::new();
+        let mut input_moves: Vec<(Reg, Loc)> = Vec::with_capacity(asm_data.inputs.len());
 
         // Track register remaps: if an allocated reg conflicts with reserved, use temp
         // (original_reg, temp_reg, actual_loc for restore)
-        let mut remap_setup: Vec<(Reg, Reg, Loc)> = Vec::new();
-        let mut remap_restore: Vec<(Reg, Reg, Loc)> = Vec::new();
+        let mut remap_setup: Vec<(Reg, Reg, Loc)> = Vec::with_capacity(operand_count);
+        let mut remap_restore: Vec<(Reg, Reg, Loc)> = Vec::with_capacity(operand_count);
 
         // Track which pseudos have been assigned temp registers (for +r where input/output share pseudo)
         let mut pseudo_to_temp: std::collections::HashMap<PseudoId, Reg> =

cc/ir/linearize.rs

@@ -26,6 +26,11 @@ use crate::target::Target;
 use crate::types::{MemberInfo, TypeId, TypeKind, TypeModifiers, TypeTable};
 use std::collections::HashMap;
 
+/// Default capacity for variable mapping HashMaps during IR linearization
+const DEFAULT_VAR_MAP_CAPACITY: usize = 64;
+/// Default capacity for label/static HashMaps (typically smaller)
+const DEFAULT_LABEL_MAP_CAPACITY: usize = 16;
+
 /// Information about a local variable
 #[derive(Clone)]
 struct LocalVarInfo {
@@ -127,9 +132,9 @@ impl<'a> Linearizer<'a> {
             current_bb: None,
             next_pseudo: 0,
             next_bb: 0,
-            var_map: HashMap::new(),
-            locals: HashMap::new(),
-            label_map: HashMap::new(),
+            var_map: HashMap::with_capacity(DEFAULT_VAR_MAP_CAPACITY),
+            locals: HashMap::with_capacity(DEFAULT_VAR_MAP_CAPACITY),
+            label_map: HashMap::with_capacity(DEFAULT_LABEL_MAP_CAPACITY),
             break_targets: Vec::new(),
             continue_targets: Vec::new(),
             run_ssa: true, // Enable SSA conversion by default
@@ -142,7 +147,7 @@ impl<'a> Linearizer<'a> {
             current_func_name: String::new(),
             static_local_counter: 0,
             compound_literal_counter: 0,
-            static_locals: HashMap::new(),
+            static_locals: HashMap::with_capacity(DEFAULT_LABEL_MAP_CAPACITY),
             current_pos: None,
             target,
             current_func_is_non_static_inline: false,
@@ -833,9 +838,11 @@ impl<'a> Linearizer<'a> {
         // Add parameters
         // For struct/union parameters, we need to copy them to local storage
         // so member access works properly
-        let mut struct_params: Vec<(String, TypeId, PseudoId)> = Vec::new();
+        let mut struct_params: Vec<(String, TypeId, PseudoId)> =
+            Vec::with_capacity(func.params.len());
         // Complex parameters also need local storage for real/imag access
-        let mut complex_params: Vec<(String, TypeId, PseudoId)> = Vec::new();
+        let mut complex_params: Vec<(String, TypeId, PseudoId)> =
+            Vec::with_capacity(func.params.len());
 
         for (i, param) in func.params.iter().enumerate() {
             let name = param

cc/symbol.rs

@@ -14,6 +14,9 @@ use crate::strings::StringId;
 use crate::types::TypeId;
 use std::collections::HashMap;
 
+/// Default capacity for symbol table name lookup HashMap
+const DEFAULT_SYMBOL_MAP_CAPACITY: usize = 256;
+
 // ============================================================================
 // Symbol ID
 // ============================================================================
@@ -234,7 +237,7 @@ impl SymbolTable {
             scopes: Vec::new(),
             current_scope: 0,
             scope_depth: 0,
-            name_map: HashMap::new(),
+            name_map: HashMap::with_capacity(DEFAULT_SYMBOL_MAP_CAPACITY),
         };
         // Create the global scope
         table.scopes.push(Scope::new(None));

cc/token/preprocess.rs

@@ -23,6 +23,9 @@ use crate::diag;
 use crate::os;
 use crate::target::Target;
 
+/// Default capacity for macro HashMap (covers predefined macros + typical program macros)
+const DEFAULT_MACRO_CAPACITY: usize = 32;
+
 // ============================================================================
 // Macro Definition
 // ============================================================================
@@ -390,7 +393,7 @@ impl<'a> Preprocessor<'a> {
 
         let mut pp = Self {
             target,
-            macros: HashMap::new(),
+            macros: HashMap::with_capacity(DEFAULT_MACRO_CAPACITY),
             cond_stack: Vec::new(),
             system_include_paths: Vec::new(),
             quote_include_paths: Vec::new(),

sh/shell/environment.rs

@@ -174,7 +174,7 @@ impl Environment {
     }
 
     pub fn exported(&self) -> impl Iterator<Item = (&String, &String)> {
-        let mut exported = HashMap::new();
+        let mut exported = HashMap::with_capacity(self.global_scope.len());
         for (name, var) in &self.global_scope {
             if var.export {
                 if let Some(value) = &var.value {

sh/shell/mod.rs

@@ -44,6 +44,9 @@ use std::rc::Rc;
 use std::time::Duration;
 use std::{env, io};
 
+/// Default capacity for cached command locations HashMap
+const DEFAULT_COMMAND_CACHE_CAPACITY: usize = 64;
+
 pub mod environment;
 pub mod history;
 pub mod opened_files;
@@ -1058,7 +1061,7 @@ impl Default for Shell {
             background_jobs: JobManager::default(),
             history: History::new(32767),
             umask: !0o022 & 0o777,
-            saved_command_locations: HashMap::new(),
+            saved_command_locations: HashMap::with_capacity(DEFAULT_COMMAND_CACHE_CAPACITY),
             is_subshell: false,
             last_pipeline_command: String::new(),
             terminal: Terminal::default(),

text/cut.rs

@@ -126,7 +126,7 @@ fn is_character_boundary(bytes: &[u8]) -> bool {
 /// A vector containing the selected bytes from the input line based on the specified ranges.
 ///
 fn cut_bytes(line: &[u8], delim: Option<char>, ranges: &Vec<(i32, i32)>, n: bool) -> Vec<u8> {
-    let mut result = Vec::new();
+    let mut result = Vec::with_capacity(line.len());
 
     for (start, end) in ranges {
         let mut start = *start as usize;

text/sed.rs

@@ -1333,7 +1333,7 @@ fn print_multiline_binary(line: &str) {
 
 /// Find first label in [`Script`] that has duplicates
 fn find_first_repeated_label(vec: Vec<String>) -> Option<String> {
-    let mut counts = HashMap::new();
+    let mut counts = HashMap::with_capacity(vec.len());
     for item in &vec {
         *counts.entry(item).or_insert(0) += 1;
     }

tree/mv.rs

@@ -302,7 +302,7 @@ fn move_files(cfg: &MvConfig, sources: &[PathBuf], target: &Path) -> Option<()>
     let mut created_files = HashSet::new();
 
     // inode of source -> target path
-    let mut inode_map = HashMap::new();
+    let mut inode_map = HashMap::with_capacity(sources.len());
 
     // Postpone deletion when moving across filesystems because it would
     // otherwise error when copying dangling hard links