Ruby EE 2011.03 Hash Collision Fix

ruby-1.8.7-hash-randomize.patch

From 4f69a748ab820c6a8bd204f94d13d970847f575c Mon Sep 17 00:00:00 2001
From: RPM Builder <[email protected]>
Date: Thu, 29 Dec 2011 19:48:52 +0000
Subject: [PATCH] Backport randomized hash patch from upstream

---
 source/inits.c                  |    4 ++
 source/random.c                 |   74 ++++++++++++++++++++++++++++++---------
 source/st.c                     |   14 +++++++-
 source/string.c                 |    7 +++-
 source/test/ruby/test_string.rb |   13 +++++++
 source/version.c                |    2 +-
 6 files changed, 94 insertions(+), 20 deletions(-)

diff --git a/source/inits.c b/source/inits.c
index 947bbbe..a0e061f 100644
--- a/source/inits.c
+++ b/source/inits.c
@@ -38,6 +38,7 @@ void Init_Precision _((void));
 void Init_sym _((void));
 void Init_process _((void));
 void Init_Random _((void));
+void Init_RandomSeed _((void));
 void Init_Range _((void));
 void Init_Regexp _((void));
 void Init_signal _((void));
@@ -46,10 +47,13 @@ void Init_Struct _((void));
 void Init_Time _((void));
 void Init_var_tables _((void));
 void Init_version _((void));
+void Init_st _((void));
 
 void
 rb_call_inits()
 {
+    Init_RandomSeed();
+    Init_st();
     Init_sym();
     Init_var_tables();
     Init_Object();
diff --git a/source/random.c b/source/random.c
index 258b0b2..790eda8 100644
--- a/source/random.c
+++ b/source/random.c
@@ -189,6 +189,7 @@ rb_genrand_real(void)
 #include <fcntl.h>
 #endif
 
+static int seed_initialized = 0;
 static VALUE saved_seed = INT2FIX(0);
 
 static VALUE
@@ -250,27 +251,22 @@ rand_init(vseed)
     return old;
 }
 
-static VALUE
-random_seed()
+#define DEFAULT_SEED_LEN (4 * sizeof(long))
+
+static void
+fill_random_seed(ptr)
+    char *ptr;
 {
     static int n = 0;
+    unsigned long *seed;
     struct timeval tv;
     int fd;
     struct stat statbuf;
+    char *buf = (char*)ptr;
 
-    int seed_len;
-    BDIGIT *digits;
-    unsigned long *seed;
-    NEWOBJ(big, struct RBignum);
-    OBJSETUP(big, rb_cBignum, T_BIGNUM);
-
-    seed_len = 4 * sizeof(long);
-    big->sign = 1;
-    big->len = seed_len / SIZEOF_BDIGITS + 1;
-    digits = big->digits = ALLOC_N(BDIGIT, big->len);
-    seed = (unsigned long *)big->digits;
+    seed = (unsigned long *)buf;
 
-    memset(digits, 0, big->len * SIZEOF_BDIGITS);
+    memset(buf, 0, DEFAULT_SEED_LEN);
 
 #ifdef S_ISCHR
     if ((fd = open("/dev/urandom", O_RDONLY
@@ -285,7 +281,7 @@ random_seed()
 #endif
             )) >= 0) {
         if (fstat(fd, &statbuf) == 0 && S_ISCHR(statbuf.st_mode)) {
-            read(fd, seed, seed_len);
+            read(fd, seed, DEFAULT_SEED_LEN);
         }
         close(fd);
     }
@@ -296,13 +292,37 @@ random_seed()
     seed[1] ^= tv.tv_sec;
     seed[2] ^= getpid() ^ (n++ << 16);
     seed[3] ^= (unsigned long)&seed;
+}
+
+static VALUE
+make_seed_value(char *ptr)
+{
+    BDIGIT *digits;
+    NEWOBJ(big, struct RBignum);
+    OBJSETUP(big, rb_cBignum, T_BIGNUM);
+
+    RBIGNUM_SET_SIGN(big, 1);
+
+    digits = ALLOC_N(char, DEFAULT_SEED_LEN);
+    RBIGNUM(big)->digits = digits;
+    RBIGNUM(big)->len = DEFAULT_SEED_LEN / SIZEOF_BDIGITS;
+
+    MEMCPY(digits, ptr, char, DEFAULT_SEED_LEN);
 
     /* set leading-zero-guard if need. */
-    digits[big->len-1] = digits[big->len-2] <= 1 ? 1 : 0;
+    digits[RBIGNUM_LEN(big)-1] = digits[RBIGNUM_LEN(big)-2] <= 1 ? 1 : 0;
 
     return rb_big_norm((VALUE)big);
 }
 
+static VALUE
+random_seed(void)
+{
+    char buf[DEFAULT_SEED_LEN];
+    fill_random_seed(buf);
+    return make_seed_value(buf);
+}
+
 /*
  *  call-seq:
  *     srand(number=0)    => old_seed
@@ -443,6 +463,9 @@ rb_f_rand(argc, argv, obj)
     long val, max;
 
     rb_scan_args(argc, argv, "01", &vmax);
+    if (!seed_initialized) {
+       rand_init(random_seed());
+    }
     switch (TYPE(vmax)) {
       case T_FLOAT:
 	if (RFLOAT(vmax)->value <= LONG_MAX && RFLOAT(vmax)->value >= LONG_MIN) {
@@ -490,10 +513,27 @@ rb_f_rand(argc, argv, obj)
     return LONG2NUM(val);
 }
 
+static char initial_seed[DEFAULT_SEED_LEN];
+
+void
+Init_RandomSeed(void)
+{
+    fill_random_seed(initial_seed);
+    init_by_array((unsigned long*)initial_seed, DEFAULT_SEED_LEN/sizeof(unsigned long));
+    seed_initialized = 1;
+}
+
+static void
+Init_RandomSeed2(void)
+{
+    saved_seed = make_seed_value(initial_seed);
+    memset(initial_seed, 0, DEFAULT_SEED_LEN);
+}
+
 void
 Init_Random()
 {
-    rand_init(random_seed());
+    Init_RandomSeed2();
     rb_define_global_function("srand", rb_f_srand, -1);
     rb_define_global_function("rand", rb_f_rand, -1);
     rb_global_variable(&saved_seed);
diff --git a/source/st.c b/source/st.c
index c16c310..21e157a 100644
--- a/source/st.c
+++ b/source/st.c
@@ -9,6 +9,7 @@
 #include <stdlib.h>
 #endif
 #include <string.h>
+#include <limits.h>
 #include "st.h"
 
 typedef struct st_table_entry st_table_entry;
@@ -521,6 +522,8 @@ st_foreach(table, func, arg)
     return 0;
 }
 
+static unsigned long hash_seed = 0;
+
 static int
 strhash(string)
     register const char *string;
@@ -550,10 +553,11 @@ strhash(string)
 
     return val + (val << 15);
 #else
-    register int val = 0;
+    register unsigned long val = hash_seed;
 
     while ((c = *string++) != '\0') {
 	val = val*997 + c;
+	val = (val << 13) | (val >> (sizeof(st_data_t) * CHAR_BIT - 13));
     }
 
     return val + (val>>5);
@@ -573,3 +577,11 @@ numhash(n)
 {
     return n;
 }
+
+extern unsigned long rb_genrand_int32(void);
+
+void
+Init_st(void)
+{
+    hash_seed = rb_genrand_int32();
+}
diff --git a/source/string.c b/source/string.c
index c6b2301..94a0281 100644
--- a/source/string.c
+++ b/source/string.c
@@ -875,13 +875,15 @@ rb_str_concat(str1, str2)
     return str1;
 }
 
+static unsigned long hash_seed;
+
 int
 rb_str_hash(str)
     VALUE str;
 {
     register long len = RSTRING(str)->len;
     register char *p = RSTRING(str)->ptr;
-    register int key = 0;
+    register unsigned long key = hash_seed;
 
 #if defined(HASH_ELFHASH)
     register unsigned int g;
@@ -905,6 +907,7 @@ rb_str_hash(str)
     while (len--) {
 	key = key*65599 + *p;
 	p++;
+	key = (key << 13) | (key >> ((sizeof(unsigned long) * CHAR_BIT) - 13));
     }
     key = key + (key>>5);
 #endif
@@ -5062,4 +5065,6 @@ Init_String()
     rb_fs = Qnil;
     rb_define_variable("$;", &rb_fs);
     rb_define_variable("$-F", &rb_fs);
+
+    hash_seed = rb_genrand_int32();
 }
diff --git a/source/test/ruby/test_string.rb b/source/test/ruby/test_string.rb
index 5f2c54f..4d97182 100644
--- a/source/test/ruby/test_string.rb
+++ b/source/test/ruby/test_string.rb
@@ -1,4 +1,5 @@
 require 'test/unit'
+require File.expand_path('envutil', File.dirname(__FILE__))
 
 class TestString < Test::Unit::TestCase
   def check_sum(str, bits=16)
@@ -29,4 +30,16 @@ class TestString < Test::Unit::TestCase
   ensure
     $KCODE = original_kcode
   end
+
+  def test_hash_random
+    str = 'abc'
+    a = [str.hash.to_s]
+    cmd = sprintf("%s -e 'print %s.hash'", EnvUtil.rubybin, str.dump)
+    3.times {
+      IO.popen(cmd, "rb") {|o|
+        a << o.read
+      }
+    }
+    assert_not_equal([str.hash.to_s], a.uniq)
+  end
 end
diff --git a/source/version.c b/source/version.c
index 8b41cc9..5781cab 100644
--- a/source/version.c
+++ b/source/version.c
@@ -46,7 +46,7 @@ Init_version()
     rb_define_global_const("RUBY_PATCHLEVEL", INT2FIX(RUBY_PATCHLEVEL));
 
     snprintf(description, sizeof(description),
-             "ruby %s (%s %s %d) [%s], MBARI 0x%x, Ruby Enterprise Edition %s",
+             "ruby %s (%s %s %d) [%s], MBARI 0x%x, Ruby Enterprise Edition %s (with hash random)",
              RUBY_VERSION, RUBY_RELEASE_DATE, RUBY_RELEASE_STR,
              RUBY_RELEASE_NUM, RUBY_PLATFORM,
              STACK_WIPE_SITES, REE_VERSION);
-- 
1.7.6.4

The patch is for 2011.03.

I saw that. I looks like it should be fine, but I must be missing something.