Simple string interning.

3 files changed, 21 insertions, 3 deletions

diff --git a/asm.peg b/asm.peg
index 0a80d3a..f181cfe 100644
--- a/asm.peg
+++ b/asm.peg
@@ -53,8 +53,8 @@ section-directive =
       {$$.section = (DirSection){.kind=ASM_DIR_SECTION, .name=n.charptr, .flags="", .type=SHT_PROGBITS}; }
   )
 
-section-name = <[.a-zA-Z0-9\-]+> { $$.charptr = xstrdup(yytext); }
-section-flags = '"' <[awx]*> '"' { $$.charptr = xstrdup(yytext); }
+section-name = <[.a-zA-Z0-9\-]+> { $$.charptr = internstring(yytext); }
+section-flags = '"' <[awx]*> '"' { $$.charptr = internstring(yytext); }
 section-type =
     "@nobits" { $$.i64 = SHT_NOBITS; }
   | "@progbits" { $$.i64 = SHT_PROGBITS; }
@@ -589,7 +589,7 @@ value =
 
 ident = 
   <[._a-zA-Z][._a-zA-Z0-9]*>
-  { $$.charptr = xstrdup(yytext); } 
+  { $$.charptr = internstring(yytext); } 
 
 number =
   '-' ws? n:unsigned-number
diff --git a/minias.h b/minias.h
index cbaae9b..8855756 100644
--- a/minias.h
+++ b/minias.h
@@ -348,6 +348,8 @@ char *xmemdup(const char *, size_t);
 char *xstrdup(const char *s);
 void *zalloc(size_t n);
 
+const char *internstring(const char *s);
+
 struct hashtable {
   size_t len, cap;
   struct hashtablekey *keys;
diff --git a/util.c b/util.c
index 9ed1c50..5cede75 100644
--- a/util.c
+++ b/util.c
@@ -76,6 +76,22 @@ char *xmemdup(const char *s, size_t n) {
 
 char *xstrdup(const char *s) { return xmemdup(s, strlen(s) + 1); }
 
+const char *internstring(const char *s) {
+  size_t idx, len;
+  const char *interned;
+  static const char *cache[4096] = {0};
+
+  len = strlen(s);
+  idx = murmurhash64a(s, len) % sizeof(cache)/sizeof(cache[0]);
+  interned = cache[idx];
+  if (interned && strcmp(s, cache[idx]) == 0) {
+    return interned;
+  }
+  interned = xstrdup(s);
+  cache[idx] = interned;
+  return interned;
+}
+
 void htabkey(struct hashtablekey *k, const char *s, size_t n) {
   k->str = s;
   k->len = n;