From 0835f40ae25f97360dc393928796387d3cd6b16e Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Wed, 19 Apr 2017 00:43:37 +0100
Subject: erts: Add enif_phash2 and enif_phash2_ranged

These allow one to hash VM terms from NIF code.
---
 erts/emulator/test/nif_SUITE.erl              | 85 ++++++++++++++++++++++++++-
 erts/emulator/test/nif_SUITE_data/nif_SUITE.c | 19 ++++++
 2 files changed, 102 insertions(+), 2 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index 693db42e58..780fe0a5ee 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -56,7 +56,9 @@
          nif_is_process_alive/1, nif_is_port_alive/1,
          nif_term_to_binary/1, nif_binary_to_term/1,
          nif_port_command/1,
-         nif_snprintf/1
+         nif_snprintf/1,
+         nif_phash2/1,
+         nif_phash2_ranged/1
 	]).
 
 -export([many_args_100/100]).
@@ -90,7 +92,9 @@ all() ->
      nif_is_process_alive, nif_is_port_alive,
      nif_term_to_binary, nif_binary_to_term,
      nif_port_command,
-     nif_snprintf].
+     nif_snprintf,
+     nif_phash2,
+     nif_phash2_ranged].
 
 groups() ->
     [{G, [], api_repeaters()} || G <- api_groups()]
@@ -2610,6 +2614,81 @@ nif_snprintf(Config) ->
     <<"{{hello,world,-33},",0>> = format_term_nif(20,{{hello,world, -33}, 3.14, self()}),
     ok.
 
+nif_phash2(Config) ->
+    ensure_lib_loaded(Config),
+    Terms =
+        [random_term() || _ <- lists:seq(1, 1000)],
+
+    lists:foreach(
+      fun (Term) ->
+              HashValue = erlang:phash2(Term),
+              NifHashValue = phash2_nif(Term),
+              (HashValue =:= NifHashValue
+               orelse ct:fail("Expected: ~p\nActual:   ~p",
+                              [HashValue, NifHashValue]))
+      end,
+      Terms).
+
+nif_phash2_ranged(Config) ->
+    ensure_lib_loaded(Config),
+    RandomRangedTerms =
+        [{random_term(), rand:uniform((1 bsl 32) - 1)}
+         || _ <- lists:seq(1, 1000)],
+
+    lists:foreach(
+      fun ({Term, Range}) ->
+              HashValue = erlang:phash2(Term, Range),
+              NifHashValue = phash2_ranged_nif(Term, Range),
+              (HashValue =:= NifHashValue
+               orelse ct:fail("Expected: ~p\nActual:   ~p",
+                              [HashValue, NifHashValue]))
+      end,
+      RandomRangedTerms),
+
+    EdgeCaseTerm = random_term(),
+    EdgeCaseHashValue = erlang:phash2(EdgeCaseTerm, 1 bsl 32),
+    EdgeCaseNifHashValue = phash2_ranged_nif(EdgeCaseTerm, 0),
+    (EdgeCaseHashValue =:= EdgeCaseNifHashValue
+     orelse ct:fail("Expected: ~p\nActual:   ~p",
+                    [EdgeCaseHashValue, EdgeCaseNifHashValue])).
+
+-define(HALF_DBL_EPSILON, 1.1102230246251565e-16). % math:pow(2, -53)
+
+random_term() ->
+    case rand:uniform(6) of
+        1 -> rand:uniform(1 bsl 27) - 1; % small
+        2 -> (1 bsl 27) + rand:uniform(1 bsl 128); % big
+        3 -> random_sign() * (rand:uniform() * ?HALF_DBL_EPSILON); % float
+        4 -> random_binary();
+        5 -> random_pid();
+        6 ->
+            Length = rand:uniform(10),
+            List = [random_term() || _ <- lists:seq(1, Length)],
+            case rand:uniform(2) of
+                1 ->
+                   List;
+                2 ->
+                   list_to_tuple(List)
+            end
+    end.
+
+random_sign() ->
+    case rand:uniform(2) of
+        1 -> -1.0;
+        2 -> 1.0
+    end.
+
+random_binary() ->
+    list_to_binary(random_bytes(rand:uniform(32) - 1)).
+
+random_bytes(0) ->
+    [];
+random_bytes(N) when N > 0 ->
+    [rand:uniform(256) - 1 | random_bytes(N - 1)].
+
+random_pid() ->
+    Processes = erlang:processes(),
+    lists:nth(rand:uniform(length(Processes)), Processes).
 
 %% The NIFs:
 lib_version() -> undefined.
@@ -2621,6 +2700,8 @@ type_test() -> ?nif_stub.
 tuple_2_list(_) -> ?nif_stub.    
 is_identical(_,_) -> ?nif_stub.
 compare(_,_) -> ?nif_stub.
+phash2_nif(_) -> ?nif_stub.
+phash2_ranged_nif(_, _) -> ?nif_stub.
 many_args_100(_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_) -> ?nif_stub.
 clone_bin(_) -> ?nif_stub.
 make_sub_bin(_,_,_) -> ?nif_stub.
diff --git a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
index 8fe5ee809a..3bdd4d93c7 100644
--- a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
+++ b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
@@ -687,6 +687,23 @@ static ERL_NIF_TERM compare(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
     return enif_make_int(env, enif_compare(argv[0],argv[1]));
 }
 
+static ERL_NIF_TERM phash2_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+    if (argc != 1) {
+        return enif_make_badarg(env);
+    }
+    return enif_make_ulong(env, enif_phash2(argv[0]));
+}
+
+static ERL_NIF_TERM phash2_ranged_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+{
+    unsigned long range;
+    if (argc != 2 || !enif_get_ulong(env, argv[1], &range)) {
+        return enif_make_badarg(env);
+    }
+    return enif_make_ulong(env, enif_phash2_ranged(argv[0], range));
+}
+
 static ERL_NIF_TERM many_args_100(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 {
     int i, k;
@@ -2864,6 +2881,8 @@ static ErlNifFunc nif_funcs[] =
     {"tuple_2_list", 1, tuple_2_list},
     {"is_identical",2,is_identical},
     {"compare",2,compare},
+    {"phash2_nif",1,phash2_nif},
+    {"phash2_ranged_nif",2,phash2_ranged_nif},
     {"many_args_100", 100, many_args_100},
     {"clone_bin", 1, clone_bin},
     {"make_sub_bin", 3, make_sub_bin},
-- 
cgit v1.2.3


From 1b37d0b010ea31b04b9d0a15d0bec9c75a013dc9 Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Thu, 20 Apr 2017 21:35:54 +0100
Subject: erts: Remove enif_phash2_ranged

---
 erts/emulator/test/nif_SUITE.erl              | 30 ++-------------------------
 erts/emulator/test/nif_SUITE_data/nif_SUITE.c | 10 ---------
 2 files changed, 2 insertions(+), 38 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index 780fe0a5ee..f868b018c9 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -57,8 +57,7 @@
          nif_term_to_binary/1, nif_binary_to_term/1,
          nif_port_command/1,
          nif_snprintf/1,
-         nif_phash2/1,
-         nif_phash2_ranged/1
+         nif_phash2/1
 	]).
 
 -export([many_args_100/100]).
@@ -93,8 +92,7 @@ all() ->
      nif_term_to_binary, nif_binary_to_term,
      nif_port_command,
      nif_snprintf,
-     nif_phash2,
-     nif_phash2_ranged].
+     nif_phash2].
 
 groups() ->
     [{G, [], api_repeaters()} || G <- api_groups()]
@@ -2629,29 +2627,6 @@ nif_phash2(Config) ->
       end,
       Terms).
 
-nif_phash2_ranged(Config) ->
-    ensure_lib_loaded(Config),
-    RandomRangedTerms =
-        [{random_term(), rand:uniform((1 bsl 32) - 1)}
-         || _ <- lists:seq(1, 1000)],
-
-    lists:foreach(
-      fun ({Term, Range}) ->
-              HashValue = erlang:phash2(Term, Range),
-              NifHashValue = phash2_ranged_nif(Term, Range),
-              (HashValue =:= NifHashValue
-               orelse ct:fail("Expected: ~p\nActual:   ~p",
-                              [HashValue, NifHashValue]))
-      end,
-      RandomRangedTerms),
-
-    EdgeCaseTerm = random_term(),
-    EdgeCaseHashValue = erlang:phash2(EdgeCaseTerm, 1 bsl 32),
-    EdgeCaseNifHashValue = phash2_ranged_nif(EdgeCaseTerm, 0),
-    (EdgeCaseHashValue =:= EdgeCaseNifHashValue
-     orelse ct:fail("Expected: ~p\nActual:   ~p",
-                    [EdgeCaseHashValue, EdgeCaseNifHashValue])).
-
 -define(HALF_DBL_EPSILON, 1.1102230246251565e-16). % math:pow(2, -53)
 
 random_term() ->
@@ -2701,7 +2676,6 @@ tuple_2_list(_) -> ?nif_stub.
 is_identical(_,_) -> ?nif_stub.
 compare(_,_) -> ?nif_stub.
 phash2_nif(_) -> ?nif_stub.
-phash2_ranged_nif(_, _) -> ?nif_stub.
 many_args_100(_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_) -> ?nif_stub.
 clone_bin(_) -> ?nif_stub.
 make_sub_bin(_,_,_) -> ?nif_stub.
diff --git a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
index 3bdd4d93c7..dac6d02e9d 100644
--- a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
+++ b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
@@ -695,15 +695,6 @@ static ERL_NIF_TERM phash2_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv
     return enif_make_ulong(env, enif_phash2(argv[0]));
 }
 
-static ERL_NIF_TERM phash2_ranged_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
-{
-    unsigned long range;
-    if (argc != 2 || !enif_get_ulong(env, argv[1], &range)) {
-        return enif_make_badarg(env);
-    }
-    return enif_make_ulong(env, enif_phash2_ranged(argv[0], range));
-}
-
 static ERL_NIF_TERM many_args_100(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 {
     int i, k;
@@ -2882,7 +2873,6 @@ static ErlNifFunc nif_funcs[] =
     {"is_identical",2,is_identical},
     {"compare",2,compare},
     {"phash2_nif",1,phash2_nif},
-    {"phash2_ranged_nif",2,phash2_ranged_nif},
     {"many_args_100", 100, many_args_100},
     {"clone_bin", 1, clone_bin},
     {"make_sub_bin", 3, make_sub_bin},
-- 
cgit v1.2.3


From 5ad95cca3f3c6802cd71fdd35dd397c3a37dc239 Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Thu, 20 Apr 2017 21:53:54 +0100
Subject: erts: Fix random floats in enif_phash2 tests

---
 erts/emulator/test/nif_SUITE.erl | 4 +---
 1 file changed, 1 insertion(+), 3 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index f868b018c9..9127a5eae9 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -2627,13 +2627,11 @@ nif_phash2(Config) ->
       end,
       Terms).
 
--define(HALF_DBL_EPSILON, 1.1102230246251565e-16). % math:pow(2, -53)
-
 random_term() ->
     case rand:uniform(6) of
         1 -> rand:uniform(1 bsl 27) - 1; % small
         2 -> (1 bsl 27) + rand:uniform(1 bsl 128); % big
-        3 -> random_sign() * (rand:uniform() * ?HALF_DBL_EPSILON); % float
+        3 -> random_sign() * (rand:uniform() * (1 bsl 53)); % float
         4 -> random_binary();
         5 -> random_pid();
         6 ->
-- 
cgit v1.2.3


From d8756f8665a42effa2f3515369058cff4441abeb Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Thu, 20 Apr 2017 22:44:04 +0100
Subject: erts: Refactor enif_phash2 into enif_hash

A more generic hashing function which can also hash terms based on
`make_internal_hash'.
---
 erts/emulator/test/nif_SUITE.erl              | 47 +++++++++++++++++++++++++--
 erts/emulator/test/nif_SUITE_data/nif_SUITE.c | 20 +++++++++---
 2 files changed, 60 insertions(+), 7 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index 9127a5eae9..e177fa4963 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -57,6 +57,7 @@
          nif_term_to_binary/1, nif_binary_to_term/1,
          nif_port_command/1,
          nif_snprintf/1,
+         nif_internal_hash/1,
          nif_phash2/1
 	]).
 
@@ -92,6 +93,7 @@ all() ->
      nif_term_to_binary, nif_binary_to_term,
      nif_port_command,
      nif_snprintf,
+     nif_internal_hash,
      nif_phash2].
 
 groups() ->
@@ -2612,15 +2614,54 @@ nif_snprintf(Config) ->
     <<"{{hello,world,-33},",0>> = format_term_nif(20,{{hello,world, -33}, 3.14, self()}),
     ok.
 
+nif_internal_hash(Config) ->
+    ensure_lib_loaded(Config),
+    Terms =
+        [random_term() || _ <- lists:seq(1, 5000)],
+
+    % Unlike the phash2 hash, in this case we
+    % have nothing to compare to, so let's try
+    % and at least make sure the distribution
+    % isn't outright wrong, statistical nuances
+    % aside.
+
+    OnesPerBit =
+        lists:foldl(
+          fun (Term, Acc) ->
+                  NifHashValue = hash_nif(internal, Term),
+                  lists:foldl(
+                    fun (BitIndex, AccB) ->
+                            BitValue = (NifHashValue band (1 bsl BitIndex)) bsr BitIndex,
+                            dict:update_counter(BitIndex, BitValue, AccB)
+                    end,
+                    Acc,
+                    lists:seq(0, 31))
+          end,
+          dict:new(),
+          Terms),
+
+    ExpectedNrOfOnes = length(Terms) div 2,
+    dict:fold(
+      fun (BitIndex, NrOfOnes, Acc) ->
+              RelativeDeviation = abs(NrOfOnes - ExpectedNrOfOnes) / ExpectedNrOfOnes,
+              (RelativeDeviation < 0.10
+               orelse ct:fail("Unreasonable deviation on number of set bits (i=~p): "
+                              "expected ~p, got ~p (relative dev. ~.3f)",
+                              [BitIndex, ExpectedNrOfOnes, NrOfOnes, RelativeDeviation])),
+              Acc
+      end,
+      ok,
+      OnesPerBit).
+
 nif_phash2(Config) ->
     ensure_lib_loaded(Config),
     Terms =
-        [random_term() || _ <- lists:seq(1, 1000)],
+        [random_term() || _ <- lists:seq(1, 5000)],
 
     lists:foreach(
       fun (Term) ->
               HashValue = erlang:phash2(Term),
-              NifHashValue = phash2_nif(Term),
+              NifHashValue = hash_nif(phash2, Term),
               (HashValue =:= NifHashValue
                orelse ct:fail("Expected: ~p\nActual:   ~p",
                               [HashValue, NifHashValue]))
@@ -2673,7 +2714,7 @@ type_test() -> ?nif_stub.
 tuple_2_list(_) -> ?nif_stub.    
 is_identical(_,_) -> ?nif_stub.
 compare(_,_) -> ?nif_stub.
-phash2_nif(_) -> ?nif_stub.
+hash_nif(_, _) -> ?nif_stub.
 many_args_100(_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_) -> ?nif_stub.
 clone_bin(_) -> ?nif_stub.
 make_sub_bin(_,_,_) -> ?nif_stub.
diff --git a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
index dac6d02e9d..9d65b9c09b 100644
--- a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
+++ b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
@@ -687,12 +687,24 @@ static ERL_NIF_TERM compare(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
     return enif_make_int(env, enif_compare(argv[0],argv[1]));
 }
 
-static ERL_NIF_TERM phash2_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
+static ERL_NIF_TERM hash_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 {
-    if (argc != 1) {
+    if (argc != 2) {
+        return enif_make_badarg(env);
+    }
+
+    ErlNifHash type;
+    if (enif_is_identical(argv[0], enif_make_atom(env, "internal"))) {
+        type = ERL_NIF_INTERNAL_HASH;
+    }
+    else if (enif_is_identical(argv[0], enif_make_atom(env, "phash2"))) {
+        type = ERL_NIF_PHASH2;
+    }
+    else {
         return enif_make_badarg(env);
     }
-    return enif_make_ulong(env, enif_phash2(argv[0]));
+
+    return enif_make_ulong(env, enif_hash(type, argv[1]));
 }
 
 static ERL_NIF_TERM many_args_100(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
@@ -2872,7 +2884,7 @@ static ErlNifFunc nif_funcs[] =
     {"tuple_2_list", 1, tuple_2_list},
     {"is_identical",2,is_identical},
     {"compare",2,compare},
-    {"phash2_nif",1,phash2_nif},
+    {"hash_nif",2,hash_nif},
     {"many_args_100", 100, many_args_100},
     {"clone_bin", 1, clone_bin},
     {"make_sub_bin", 3, make_sub_bin},
-- 
cgit v1.2.3


From 21e1d879faaae2278a68200fe1085d7e73791fa0 Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Sat, 22 Apr 2017 17:22:32 +0100
Subject: erts: Support custom salt in enif_hash

---
 erts/emulator/test/nif_SUITE.erl              |  6 +++---
 erts/emulator/test/nif_SUITE_data/nif_SUITE.c | 11 ++++++++---
 2 files changed, 11 insertions(+), 6 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index e177fa4963..36af68be60 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -2628,7 +2628,7 @@ nif_internal_hash(Config) ->
     OnesPerBit =
         lists:foldl(
           fun (Term, Acc) ->
-                  NifHashValue = hash_nif(internal, Term),
+                  NifHashValue = hash_nif(internal, Term, 0),
                   lists:foldl(
                     fun (BitIndex, AccB) ->
                             BitValue = (NifHashValue band (1 bsl BitIndex)) bsr BitIndex,
@@ -2661,7 +2661,7 @@ nif_phash2(Config) ->
     lists:foreach(
       fun (Term) ->
               HashValue = erlang:phash2(Term),
-              NifHashValue = hash_nif(phash2, Term),
+              NifHashValue = hash_nif(phash2, Term, 0),
               (HashValue =:= NifHashValue
                orelse ct:fail("Expected: ~p\nActual:   ~p",
                               [HashValue, NifHashValue]))
@@ -2714,7 +2714,7 @@ type_test() -> ?nif_stub.
 tuple_2_list(_) -> ?nif_stub.    
 is_identical(_,_) -> ?nif_stub.
 compare(_,_) -> ?nif_stub.
-hash_nif(_, _) -> ?nif_stub.
+hash_nif(_Type, _Term, _Salt) -> ?nif_stub.
 many_args_100(_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_,_) -> ?nif_stub.
 clone_bin(_) -> ?nif_stub.
 make_sub_bin(_,_,_) -> ?nif_stub.
diff --git a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
index 9d65b9c09b..2a80d48b62 100644
--- a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
+++ b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
@@ -689,7 +689,7 @@ static ERL_NIF_TERM compare(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 
 static ERL_NIF_TERM hash_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
 {
-    if (argc != 2) {
+    if (argc != 3) {
         return enif_make_badarg(env);
     }
 
@@ -704,7 +704,12 @@ static ERL_NIF_TERM hash_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]
         return enif_make_badarg(env);
     }
 
-    return enif_make_ulong(env, enif_hash(type, argv[1]));
+    unsigned long salt;
+    if (! enif_get_ulong(env, argv[2], &salt)) {
+        return enif_make_badarg(env);
+    }
+
+    return enif_make_ulong(env, enif_hash(type, argv[1], salt));
 }
 
 static ERL_NIF_TERM many_args_100(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
@@ -2884,7 +2889,7 @@ static ErlNifFunc nif_funcs[] =
     {"tuple_2_list", 1, tuple_2_list},
     {"is_identical",2,is_identical},
     {"compare",2,compare},
-    {"hash_nif",2,hash_nif},
+    {"hash_nif",3,hash_nif},
     {"many_args_100", 100, many_args_100},
     {"clone_bin", 1, clone_bin},
     {"make_sub_bin", 3, make_sub_bin},
-- 
cgit v1.2.3


From c94506ccc003016db49fa2dbe5741196e071d45a Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Sat, 22 Apr 2017 18:40:26 +0100
Subject: erts: Add test cases for salted enif_hash calls

---
 erts/emulator/test/nif_SUITE.erl | 148 ++++++++++++++++++++++++++++-----------
 1 file changed, 107 insertions(+), 41 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index 36af68be60..6e4866bfd8 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -58,7 +58,9 @@
          nif_port_command/1,
          nif_snprintf/1,
          nif_internal_hash/1,
-         nif_phash2/1
+         nif_internal_hash_salted/1,
+         nif_phash2/1,
+         nif_phash2_salted/1
 	]).
 
 -export([many_args_100/100]).
@@ -94,7 +96,9 @@ all() ->
      nif_port_command,
      nif_snprintf,
      nif_internal_hash,
-     nif_phash2].
+     nif_internal_hash_salted,
+     nif_phash2,
+     nif_phash2_salted].
 
 groups() ->
     [{G, [], api_repeaters()} || G <- api_groups()]
@@ -2616,57 +2620,119 @@ nif_snprintf(Config) ->
 
 nif_internal_hash(Config) ->
     ensure_lib_loaded(Config),
-    Terms =
-        [random_term() || _ <- lists:seq(1, 5000)],
+    HashValueBitSize = nif_hash_result_bitsize(internal),
+    Terms = unique([random_term() || _ <- lists:seq(1, 5000)]),
+    HashValues = [hash_nif(internal, Term, 0) || Term <- Terms],
+    test_bit_distribution_fitness(HashValues, HashValueBitSize, 0.05).
 
-    % Unlike the phash2 hash, in this case we
-    % have nothing to compare to, so let's try
-    % and at least make sure the distribution
-    % isn't outright wrong, statistical nuances
-    % aside.
+nif_internal_hash_salted(Config) ->
+    ensure_lib_loaded(Config),
+    test_salted_nif_hash(internal).
+
+nif_phash2(Config) ->
+    ensure_lib_loaded(Config),
+    HashValueBitSize = nif_hash_result_bitsize(phash2),
+    Terms = unique([random_term() || _ <- lists:seq(1, 5000)]),
+    HashValues =
+        lists:map(
+          fun (Term) ->
+                  HashValue = erlang:phash2(Term),
+                  NifHashValue = hash_nif(phash2, Term, 0),
+                  (HashValue =:= NifHashValue
+                   orelse ct:fail("Expected: ~p\nActual:   ~p",
+                                  [HashValue, NifHashValue])),
+                  HashValue
+          end,
+          Terms),
+    test_bit_distribution_fitness(HashValues, HashValueBitSize, 0.05).
+
+nif_phash2_salted(Config) ->
+    ensure_lib_loaded(Config),
+    test_salted_nif_hash(phash2).
+
+test_salted_nif_hash(HashType) ->
+    HashValueBitSize = nif_hash_result_bitsize(HashType),
+    Terms = unique([random_term() || _ <- lists:seq(1, 5000)]),
+    Salts = unique([random_uint32() || _ <- lists:seq(1, 100)]),
+    {HashValuesPerSalt, HashValuesPerTerm} =
+        lists:mapfoldl(
+          fun (Salt, Acc) ->
+                  {HashValues, NewAcc} =
+                    lists:mapfoldl(
+                      fun (Term, AccB) ->
+                              HashValue = hash_nif(HashType, Term, Salt),
+                              NewAccB = dict:append(Term, HashValue, AccB),
+                              {HashValue, NewAccB}
+                      end,
+                      Acc,
+                      Terms),
+                  {{Salt, HashValues}, NewAcc}
+          end,
+          dict:new(),
+          Salts),
+
+    % Test per-salt hash distribution of different terms
+    lists:foreach(
+      fun ({_Salt, HashValues}) ->
+              test_bit_distribution_fitness(HashValues, HashValueBitSize, 0.05)
+      end,
+      HashValuesPerSalt),
 
+    % Test per-term hash distribution of different salts
+    dict:fold(
+      fun (_Term, HashValues, Acc) ->
+              % Be more tolerant of relative deviation,
+              % as there's fewer hash values here.
+              test_bit_distribution_fitness(HashValues, HashValueBitSize, 0.30),
+              Acc
+      end,
+      ok,
+      HashValuesPerTerm).
+
+test_bit_distribution_fitness(Integers, BitSize, MaxRelativeDeviation) ->
+    MaxInteger = (1 bsl BitSize) - 1,
     OnesPerBit =
         lists:foldl(
-          fun (Term, Acc) ->
-                  NifHashValue = hash_nif(internal, Term, 0),
+          fun (Integer, Acc) when Integer >= 0, Integer =< MaxInteger ->
                   lists:foldl(
                     fun (BitIndex, AccB) ->
-                            BitValue = (NifHashValue band (1 bsl BitIndex)) bsr BitIndex,
-                            dict:update_counter(BitIndex, BitValue, AccB)
+                            BitValue = (Integer band (1 bsl BitIndex)) bsr BitIndex,
+                            orddict:update_counter(BitIndex, BitValue, AccB)
                     end,
                     Acc,
-                    lists:seq(0, 31))
+                    lists:seq(0, BitSize - 1))
           end,
-          dict:new(),
-          Terms),
+          orddict:new(),
+          Integers),
+
+    ExpectedNrOfOnes = length(Integers) div 2,
+    FailureText =
+        orddict:fold(
+          fun (BitIndex, NrOfOnes, Acc) ->
+                  RelativeDeviation = abs(NrOfOnes - ExpectedNrOfOnes) / length(Integers),
+                  case RelativeDeviation >= MaxRelativeDeviation of
+                      false -> Acc;
+                      true ->
+                          [Acc,
+                           io_lib:format(
+                             "Unreasonable deviation on number of set bits (i=~p): "
+                             "expected ~p, got ~p (relative dev. ~.3f)~n",
+                             [BitIndex, ExpectedNrOfOnes, NrOfOnes, RelativeDeviation])]
+                  end
+          end,
+          [],
+          OnesPerBit),
 
-    ExpectedNrOfOnes = length(Terms) div 2,
-    dict:fold(
-      fun (BitIndex, NrOfOnes, Acc) ->
-              RelativeDeviation = abs(NrOfOnes - ExpectedNrOfOnes) / ExpectedNrOfOnes,
-              (RelativeDeviation < 0.10
-               orelse ct:fail("Unreasonable deviation on number of set bits (i=~p): "
-                              "expected ~p, got ~p (relative dev. ~.3f)",
-                              [BitIndex, ExpectedNrOfOnes, NrOfOnes, RelativeDeviation])),
-              Acc
-      end,
-      ok,
-      OnesPerBit).
+    (FailureText =:= [] orelse ct:fail(FailureText)).
 
-nif_phash2(Config) ->
-    ensure_lib_loaded(Config),
-    Terms =
-        [random_term() || _ <- lists:seq(1, 5000)],
+nif_hash_result_bitsize(internal) -> 32;
+nif_hash_result_bitsize(phash2) -> 27.
 
-    lists:foreach(
-      fun (Term) ->
-              HashValue = erlang:phash2(Term),
-              NifHashValue = hash_nif(phash2, Term, 0),
-              (HashValue =:= NifHashValue
-               orelse ct:fail("Expected: ~p\nActual:   ~p",
-                              [HashValue, NifHashValue]))
-      end,
-      Terms).
+unique(List) ->
+    lists:usort(List).
+
+random_uint32() ->
+    rand:uniform(1 bsl 32) - 1.
 
 random_term() ->
     case rand:uniform(6) of
-- 
cgit v1.2.3


From da9abd24a93ae8fe174cdd38fc9699bbc45fdf56 Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Sat, 22 Apr 2017 20:00:28 +0100
Subject: erts: Allow for easier future enif_hash expansion

Allow for expanding support to 64-bit hashes without breaking the
interface.
---
 erts/emulator/test/nif_SUITE_data/nif_SUITE.c | 6 +++---
 1 file changed, 3 insertions(+), 3 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
index 2a80d48b62..a255c9f096 100644
--- a/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
+++ b/erts/emulator/test/nif_SUITE_data/nif_SUITE.c
@@ -704,12 +704,12 @@ static ERL_NIF_TERM hash_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]
         return enif_make_badarg(env);
     }
 
-    unsigned long salt;
-    if (! enif_get_ulong(env, argv[2], &salt)) {
+    ErlNifUInt64 salt;
+    if (! enif_get_uint64(env, argv[2], &salt)) {
         return enif_make_badarg(env);
     }
 
-    return enif_make_ulong(env, enif_hash(type, argv[1], salt));
+    return enif_make_uint64(env, enif_hash(type, argv[1], salt));
 }
 
 static ERL_NIF_TERM many_args_100(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[])
-- 
cgit v1.2.3


From 3d6b8e7fb68543713f1620a45b8a590ef4ed88a5 Mon Sep 17 00:00:00 2001
From: Guilherme Andrade <g@gandrade.net>
Date: Mon, 24 Apr 2017 22:51:49 +0100
Subject: erts: Discontinue salted use of enif_hash/phash2

---
 erts/emulator/test/nif_SUITE.erl | 13 ++++---------
 1 file changed, 4 insertions(+), 9 deletions(-)

(limited to 'erts/emulator/test')

diff --git a/erts/emulator/test/nif_SUITE.erl b/erts/emulator/test/nif_SUITE.erl
index 6e4866bfd8..8ad11d3bf3 100644
--- a/erts/emulator/test/nif_SUITE.erl
+++ b/erts/emulator/test/nif_SUITE.erl
@@ -59,8 +59,7 @@
          nif_snprintf/1,
          nif_internal_hash/1,
          nif_internal_hash_salted/1,
-         nif_phash2/1,
-         nif_phash2_salted/1
+         nif_phash2/1
 	]).
 
 -export([many_args_100/100]).
@@ -97,8 +96,7 @@ all() ->
      nif_snprintf,
      nif_internal_hash,
      nif_internal_hash_salted,
-     nif_phash2,
-     nif_phash2_salted].
+     nif_phash2].
 
 groups() ->
     [{G, [], api_repeaters()} || G <- api_groups()]
@@ -2637,7 +2635,8 @@ nif_phash2(Config) ->
         lists:map(
           fun (Term) ->
                   HashValue = erlang:phash2(Term),
-                  NifHashValue = hash_nif(phash2, Term, 0),
+                  Salt = random_uint32(), % phash2 should ignore salt
+                  NifHashValue = hash_nif(phash2, Term, Salt),
                   (HashValue =:= NifHashValue
                    orelse ct:fail("Expected: ~p\nActual:   ~p",
                                   [HashValue, NifHashValue])),
@@ -2646,10 +2645,6 @@ nif_phash2(Config) ->
           Terms),
     test_bit_distribution_fitness(HashValues, HashValueBitSize, 0.05).
 
-nif_phash2_salted(Config) ->
-    ensure_lib_loaded(Config),
-    test_salted_nif_hash(phash2).
-
 test_salted_nif_hash(HashType) ->
     HashValueBitSize = nif_hash_result_bitsize(HashType),
     Terms = unique([random_term() || _ <- lists:seq(1, 5000)]),
-- 
cgit v1.2.3