ruby-dev

なかだです。

At Wed, 12 Mar 2008 19:50:18 +0900,
NARUSE, Yui wrote in [ruby-dev:34026]:
> というわけで、int32_t, uint32_t, intptr_t, uintptr_t を定義しつつ、
> hash() をアライメント考慮させるパッチです。

inttypes.hをチェックしてないような気がします。

あとBig endianになるとやや遅くなりそうなので、一応作っておいたの
も出してみます。


Index: configure.in
===================================================================
--- configure.in	(revision 15758)
+++ configure.in	(working copy)
@@ -583,5 +583,5 @@ AC_CHECK_HEADERS(stdlib.h string.h unist
 		 syscall.h pwd.h grp.h a.out.h utime.h memory.h direct.h sys/resource.h \
 		 sys/mkdev.h sys/utime.h xti.h netinet/in_systm.h float.h ieeefp.h pthread.h \
-		 ucontext.h intrinsics.h langinfo.h locale.h)
+		 ucontext.h intrinsics.h langinfo.h locale.h stdint.h)
 
 dnl Check additional types.
@@ -621,4 +621,16 @@ AC_CHECK_TYPES(struct timespec)
 AC_CHECK_TYPE(fd_mask, [AC_DEFINE(HAVE_RB_FD_INIT, 1)])
 
+test ${rb_cv_type_uint32_t+set} && ac_cv_type_uint32_t=yes
+AC_CHECK_TYPE(uint32_t)
+if test ${ac_cv_type_uint32_t} != yes; then
+    AC_CACHE_CHECK([unsigned 32bit int],
+	rb_cv_type_uint32_t,
+	[for type in short int long; do
+	    AC_COMPILE_IFELSE(AC_LANG_BOOL_COMPILE_TRY([], [sizeof($type) == 4]),
+		[rb_cv_type_uint32_t=$type; break], [])
+	done])
+    AC_DEFINE(uint32_t, $rb_cv_type_uint32_t)
+fi
+
 AC_CACHE_CHECK(for stack end address, rb_cv_stack_end_address,
 [rb_cv_stack_end_address=no
Index: string.c
===================================================================
--- string.c	(revision 15758)
+++ string.c	(working copy)
@@ -26,4 +26,8 @@
 #endif
 
+#if HAVE_STDINT_H
+#include <stdint.h>
+#endif
+
 VALUE rb_cString;
 VALUE rb_cSymbol;
@@ -1686,4 +1690,11 @@ rb_str_concat(VALUE str1, VALUE str2)
 }
 
+#if defined __i386__ || defined _M_IX86 || defined __ia64
+#define UNALIGNED_WORD_ACCESS 1
+#endif
+#ifndef UNALIGNED_WORD_ACCESS
+#define UNALIGNED_WORD_ACCESS 0
+#endif
+
 /* MurmurHash described in http://murmurhash.googlepages.com/ */
 unsigned int
@@ -1695,14 +1706,93 @@ hash(const unsigned char * data, int len
     h += 0xdeadbeef;
 
-    while(len >= 4) {
-	h += *(unsigned int *)data;
-	h *= m;
-	h ^= h >> r;
+    if (len >= 4) {
+#if !UNALIGNED_WORD_ACCESS
+	int align = (VALUE)data & 3;
+	if (align) {
+	    uint32_t t = 0, d = 0;
+	    int sl, sr, pack;
+
+	    switch (align) {
+#ifdef WORDS_BIGENDIAN
+	      case 1: t |= data[2];
+	      case 2: t |= data[1] << 8;
+	      case 3: t |= data[0] << 16;
+#else
+	      case 1: t |= data[2] << 16;
+	      case 2: t |= data[1] << 8;
+	      case 3: t |= data[0];
+#endif
+	    }
+
+	    t <<= (8 * align);
+
+	    data += 4-align;
+	    len -= 4-align;
+
+	    sl = 8 * (4-align);
+	    sr = 8 * align;
+
+	    while (len >= 4) {
+		d = *(uint32_t *)data;
+#ifdef WORDS_BIGENDIAN
+		t = (t << sr) | (d >> sl);
+#else
+		t = (t >> sr) | (d << sl);
+#endif
+		h += t;
+		h *= m;
+		h ^= h >> r;
+		t = d;
+
+		data += 4;
+		len -= 4;
+	    }
+
+	    pack = len < align ? len : align;
+	    d = 0;
+	    switch (pack) {
+#ifdef WORDS_BIGENDIAN
+	      case 3: d |= data[2];
+	      case 2: d |= data[1] << 8;
+	      case 1: d |= data[0] << 16;
+	      case 0:
+		h += (t << sr) | (d >> sl);
+#else
+	      case 3: d |= data[2] << 16;
+	      case 2: d |= data[1] << 8;
+	      case 1: d |= data[0];
+	      case 0:
+		h += (t >> sr) | (d << sl);
+#endif
+		h *= m;
+		h ^= h >> r;
+	    }
 
-	data += 4;
-	len -= 4;
+	    data += pack;
+	    len -= pack;
+	}
+	else
+#endif
+	{
+	    do {
+		h += *(uint32_t *)data;
+		h *= m;
+		h ^= h >> r;
+
+		data += 4;
+		len -= 4;
+	    } while (len >= 4);
+	}
     }
 
     switch(len) {
+#ifdef WORDS_BIGENDIAN
+      case 3:
+	h += data[2];
+      case 2:
+	h += data[1] << 8;
+      case 1:
+	h += data[0] << 16;
+#else
       case 3:
 	h += data[2] << 16;
@@ -1711,4 +1801,5 @@ hash(const unsigned char * data, int len
       case 1:
 	h += data[0];
+#endif
 	h *= m;
 	h ^= h >> r;


-- 
--- 僕の前にBugはない。
--- 僕の後ろにBugはできる。
    中田 伸悦