Implement universal hashing for niecza hashes

It appears to be a few percent slower than the builtin string.GetHashCode, but proof against hash-based algorithmic complexity attacks is hard to replace...
sorear · Mar 30, 2012 · e08dd76 · e08dd76
1 parent b4b916a
commit e08dd76
Show file tree

Hide file tree

Showing 3 changed files with 46 additions and 19 deletions.
diff --git a/lib/CodeGen.cs b/lib/CodeGen.cs
@@ -3652,7 +3652,7 @@ public class DowncallReceiver : CallReceiver {
                 Kernel.TraceFlags = Kernel.TRACE_CUR;
                 Kernel.TraceCount = Kernel.TraceFreq = 1;
             }
-            Kernel.SetTrace();
+            Kernel.InitGlobal();
             Backend.obj_dir = (string)args[1];
             Builtins.upcall_receiver = (System.Collections.IDictionary)args[2];
             return null;

diff --git a/lib/Kernel.cs b/lib/Kernel.cs
@@ -4740,6 +4740,15 @@ public class Kernel {
         [CompartmentGlobal] internal static SubInfo CommonGrep_I;
         [CompartmentGlobal] internal static SubInfo TEMP_SI;
 
+        internal static void InitGlobal() {
+            Random r = new Random();
+            VarHash.string_hash_argument =
+                (uint)r.Next(VarHash.HASH_ARG_MAX + 1);
+            VarHash.hash_automorphism =
+                ((uint)r.Next(int.MaxValue)) * 2 + 1;
+            SetTrace();
+        }
+
         internal static void InitCompartment() {
             RuntimeUnit.reg = new ObjectRegistry();
 
@@ -6518,7 +6527,7 @@ class LastFrameNode {
 
         public static void MainHandler(string uname, string[] args) {
             InitCompartment();
-            SetTrace();
+            InitGlobal();
             commandArgs = args;
 
             RuntimeUnit ru = (RuntimeUnit)
@@ -6555,7 +6564,7 @@ class LastFrameNode {
             string cmd = args.Length > 0 ? args[0] : "-help";
 
             InitCompartment();
-            SetTrace();
+            InitGlobal();
 
             if (cmd == "-field-inventory") {
                 foreach (Type ty in typeof(Kernel).Assembly.GetTypes()) {

diff --git a/lib/Utils.cs b/lib/Utils.cs
@@ -204,16 +204,35 @@ public sealed class VarHash : IEnumerable<KeyValuePair<string,Variable>>,
         VarHashLink[] heap;
         int[] htab;
 
-        const int INITIAL = 5;
-        const int THRESHOLD = 11;
-
-        [Immutable]
-        static int[] grow = new int[] {
-            5, 11, 17, 37, 67, 131, 257, 521, 1031, 2053, 4099, 8209, 16411,
-            32771, 65537, 131101, 262147, 524309, 1048583, 2097169, 4194319,
-            8388617, 16777259, 33554467, 67108879, 134217757, 268435459,
-            536870923, 1073741827
-        };
+        const int INITIAL = 4;
+        const int THRESHOLD = 8;
+
+        // sacrifices a bit of universality to save a reduction step
+        public const int HASH_ARG_MAX = 1073709057;
+        [TrueGlobal]
+        public static uint string_hash_argument;
+        public static uint hash_automorphism;
+
+        public static unsafe int UniversalHash(int buckets, string str) {
+            fixed(char *c = str) {
+                char *cc = c;
+                char *end = cc + str.Length;
+                uint accum = 0;
+                while (cc < end) {
+                    // accum <= 2^32-1-65535
+                    accum += *(cc++);
+                    // accum <= 2^32-1
+                    ulong temp = (ulong)accum * string_hash_argument;
+                    // temp <= (2^32-1) * HASH_ARG_MAX
+                    // temp <= 4611545284160290815
+                    accum = (uint)((temp & 0x7FFFFFFF) + (temp >> 31));
+                    // accum <= floor(temp / 2^31) + (2^31-1)
+                    // accum <= 4294901760 = 2^32-1-65535
+                }
+                ulong temp2 = (ulong)accum * (uint)buckets;
+                return (int) (temp2 >> 32);
+            }
+        }
 
         public VarHash() { Clear(); }
 
@@ -255,8 +274,7 @@ public sealed class VarHash : IEnumerable<KeyValuePair<string,Variable>>,
                     return;
                 }
 
-                int bkt = (int)(((uint) key.GetHashCode()) %
-                        ((uint) htab.Length));
+                int bkt = UniversalHash(htab.Length, key);
                 int ptr = htab[bkt];
 
                 if (ptr < 0) {
@@ -307,11 +325,11 @@ public sealed class VarHash : IEnumerable<KeyValuePair<string,Variable>>,
 
         void rehash(int ordel) {
             int rank = 0;
-            while (heap.Length != grow[rank]) rank++;
+            while (heap.Length != (1 << (rank + 2))) rank++;
             rank += ordel;
 
             VarHashLink[] oheap = heap;
-            init(grow[rank]);
+            init(1 << (rank + 2));
 
             foreach (VarHashLink vhl in oheap)
                 if (vhl.key != null)
@@ -338,7 +356,7 @@ public sealed class VarHash : IEnumerable<KeyValuePair<string,Variable>>,
                 return false;
             }
 
-            int bkt = (int)(((uint) key.GetHashCode()) % ((uint) htab.Length));
+            int bkt = UniversalHash(htab.Length, key);
             int ptr = htab[bkt];
 
             if (ptr < 0)
@@ -419,7 +437,7 @@ public sealed class VarHash : IEnumerable<KeyValuePair<string,Variable>>,
                 return false;
             }
 
-            int ptr = htab[((uint) key.GetHashCode()) % ((uint) htab.Length)];
+            int ptr = htab[UniversalHash(htab.Length, key)];
 
             while (ptr >= 0) {
                 if (heap[ptr].key == key) {