Fix Issue 21006 - core.internal.hash.bytesHash: in 64-bit builds use a 64-bit-oriented algorithm

Author: n8sh

src/core/internal/hash.d

@@ -623,14 +623,16 @@ size_t bytesHash()(scope const(void)* buf, size_t len, size_t seed)
 
 private template bytesHashAlignedBy(AlignType)
 {
-    alias bytesHashAlignedBy = bytesHash!(AlignType.alignof >= uint.alignof);
+    static if (size_t.sizeof == 4)
+        alias bytesHashAlignedBy = bytesHash32!(AlignType.alignof >= uint.alignof);
+    else
+        alias bytesHashAlignedBy = bytesHash64!();
 }
 
 private template bytesHashWithExactSizeAndAlignment(SizeAndAlignType)
 {
-    static if (SizeAndAlignType.alignof < uint.alignof
-            ? SizeAndAlignType.sizeof <= 12
-            : SizeAndAlignType.sizeof <= 10)
+    static if (SizeAndAlignType.sizeof <= 3 || size_t.sizeof <= 4 &&
+             SizeAndAlignType.sizeof <= (SizeAndAlignType.alignof < uint.alignof ? 12 : 10))
         alias bytesHashWithExactSizeAndAlignment = smallBytesHash;
     else
         alias bytesHashWithExactSizeAndAlignment = bytesHashAlignedBy!SizeAndAlignType;
@@ -670,9 +672,11 @@ private uint get32bits()(scope const(ubyte)* x) @nogc nothrow pure @system
 Params:
     dataKnownToBeAligned = whether the data is known at compile time to be uint-aligned.
 +/
+static if (size_t.sizeof == 4)
 @nogc nothrow pure @trusted
-private size_t bytesHash(bool dataKnownToBeAligned)(scope const(ubyte)[] bytes, size_t seed)
+private uint bytesHash32(bool dataKnownToBeAligned)(scope const(ubyte)[] bytes, size_t seed)
 {
+    // MurmurHash3_x86_32.
     auto len = bytes.length;
     auto data = bytes.ptr;
     auto nblocks = len / 4;
@@ -725,6 +729,85 @@ private size_t bytesHash(bool dataKnownToBeAligned)(scope const(ubyte)[] bytes,
     return h1;
 }
 
+static if (size_t.sizeof == 8)
+@nogc nothrow pure @trusted
+private ulong bytesHash64()(scope const ubyte[] bytes, ulong seed)
+{
+    // MurmurHash3_x86_32 modified to be 64-bit using constants from MurmurHash3_x64_128.
+    alias h1 = seed;
+
+    enum ulong c1 = 0x87c37b91114253d5;
+    enum ulong c2 = 0x4cf5ad432745937f;
+    enum uint  c3 = 0x52dce729;
+
+    const(ubyte)* data = bytes.ptr;
+    //----------
+    // body
+    for (const end_data = bytes.ptr + (bytes.length & ~7);
+         data !is end_data;
+         data += 8)
+    {
+        version (BigEndian)
+        {
+            auto k1 =
+                   ((cast(ulong) data[0]) << 56) |
+                   ((cast(ulong) data[1]) << 48) |
+                   ((cast(ulong) data[2]) << 40) |
+                   ((cast(ulong) data[3]) << 32) |
+                   ((cast(ulong) data[4]) << 24) |
+                   ((cast(ulong) data[5]) << 16) |
+                   ((cast(ulong) data[6]) << 8)  |
+                    (cast(ulong) data[7]);
+        }
+        else
+        {
+            auto k1 =
+                   ((cast(ulong) data[7]) << 56) |
+                   ((cast(ulong) data[6]) << 48) |
+                   ((cast(ulong) data[5]) << 40) |
+                   ((cast(ulong) data[4]) << 32) |
+                   ((cast(ulong) data[3]) << 24) |
+                   ((cast(ulong) data[2]) << 16) |
+                   ((cast(ulong) data[1]) << 8)  |
+                    (cast(ulong) data[0]);
+        }
+
+        k1 *= c1;
+        k1 = (k1 << 31) | (k1 >> (64 - 31));
+        k1 *= c2;
+
+        h1 ^= k1;
+        h1 = (h1 << 27) | (h1 >> (64 - 27));
+        h1 = h1*5+c3;
+    }
+
+    //----------
+    // tail
+    ulong k1 = 0;
+
+    switch (bytes.length & 7)
+    {
+    case 7: k1 ^= (cast(ulong) data[6]) << 48; goto case;
+    case 6: k1 ^= (cast(ulong) data[5]) << 40; goto case;
+    case 5: k1 ^= (cast(ulong) data[4]) << 32; goto case;
+    case 4: k1 ^= (cast(ulong) data[3]) << 24; goto case;
+    case 3: k1 ^= (cast(ulong) data[2]) << 16; goto case;
+    case 2: k1 ^= (cast(ulong) data[1]) <<  8; goto case;
+    case 1: k1 ^= (cast(ulong) data[0]);
+            k1 *= c1; k1 = (k1 << 31) | (k1 >> (64 - 31)); k1 *= c2; h1 ^= k1;
+            goto default;
+    default:
+    }
+
+    //----------
+    // finalization
+    h1 ^= bytes.length;
+    // Force all bits of the hash block to avalanche.
+    h1 = (h1 ^ (h1 >> 33)) * 0xff51afd7ed558ccd;
+    h1 = (h1 ^ (h1 >> 33)) * 0xc4ceb9fe1a85ec53;
+    return h1 ^= h1 >> 33;
+}
+
 //  Check that bytesHash works with CTFE
 pure nothrow @system @nogc unittest
 {
@@ -750,7 +833,8 @@ pure nothrow @system @nogc unittest
     }
     // It is okay to change the below values if you make a change
     // that you expect to change the result of bytesHash.
-    assert(bytesHash(&a[1], a.length - 2, 0) == 2727459272);
-    assert(bytesHash(&b, 5, 0) == 2727459272);
-    assert(bytesHashAlignedBy!uint((cast(const ubyte*) &b)[0 .. 5], 0) == 2727459272);
+    enum expectedResult = size_t.sizeof == 4 ? 2727459272U : 10677742034643552556UL;
+    assert(bytesHash(&a[1], a.length - 2, 0) == expectedResult);
+    assert(bytesHash(&b, 5, 0) == expectedResult);
+    assert(bytesHashAlignedBy!uint((cast(const ubyte*) &b)[0 .. 5], 0) == expectedResult);
 }