From 13497e652a9b9495fbefd52396620bfccc6d7a7c Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?John=20H=C3=B6gberg?= <john@erlang.org>
Date: Tue, 13 Feb 2018 12:34:08 +0100
Subject: Handle unaligned binaries in erlang:iolist_to_iovec/1

A binary is a binary as long as its size in bits is evenly divisible
by 8, regardless of whether it has a bit offset or not.
---
 erts/emulator/beam/erl_io_queue.c  | 31 +++++++++++++++++++++++--------
 erts/emulator/test/iovec_SUITE.erl | 21 +++++++++++++++++++--
 2 files changed, 42 insertions(+), 10 deletions(-)

(limited to 'erts')

diff --git a/erts/emulator/beam/erl_io_queue.c b/erts/emulator/beam/erl_io_queue.c
index 190ba6bbb9..990c0c1cac 100644
--- a/erts/emulator/beam/erl_io_queue.c
+++ b/erts/emulator/beam/erl_io_queue.c
@@ -801,12 +801,11 @@ static Eterm iol2v_make_sub_bin(iol2v_state_t *state, Eterm bin_term,
     ERTS_GET_REAL_BIN(bin_term, orig_pb_term,
         byte_offset, bit_offset, bit_size);
 
-    (void)bit_offset;
-    (void)bit_size;
+    ASSERT(bit_size == 0);
 
     sb->thing_word = HEADER_SUB_BIN;
+    sb->bitoffs = bit_offset;
     sb->bitsize = 0;
-    sb->bitoffs = 0;
     sb->orig = orig_pb_term;
     sb->is_writable = 0;
 
@@ -984,7 +983,7 @@ static int iol2v_append_binary(iol2v_state_t *state, Eterm bin_term) {
     parent_header = binary_val(parent_binary);
     binary_size = binary_size(bin_term);
 
-    if (bit_offset != 0 || bit_size != 0) {
+    if (bit_size != 0) {
         return 0;
     } else if (binary_size == 0) {
         state->bytereds_spent += 1;
@@ -1026,8 +1025,16 @@ static int iol2v_append_binary(iol2v_state_t *state, Eterm bin_term) {
          * then just copy it into the accumulator. */
         iol2v_expand_acc(state, binary_size);
 
-        sys_memcpy(&(state->acc)->orig_bytes[state->acc_size],
-            binary_data, binary_size);
+        if (ERTS_LIKELY(bit_offset == 0)) {
+            sys_memcpy(&(state->acc)->orig_bytes[state->acc_size],
+                binary_data, binary_size);
+        } else {
+            ASSERT(binary_size <= ERTS_UWORD_MAX / 8);
+
+            erts_copy_bits(binary_data, bit_offset, 1,
+                (byte*)&(state->acc)->orig_bytes[state->acc_size], 0, 1,
+                binary_size * 8);
+        }
 
         state->acc_size += binary_size;
     } else {
@@ -1038,8 +1045,16 @@ static int iol2v_append_binary(iol2v_state_t *state, Eterm bin_term) {
 
         iol2v_expand_acc(state, spill);
 
-        sys_memcpy(&(state->acc)->orig_bytes[state->acc_size],
-            binary_data, spill);
+        if (ERTS_LIKELY(bit_offset == 0)) {
+            sys_memcpy(&(state->acc)->orig_bytes[state->acc_size],
+                binary_data, spill);
+        } else {
+            ASSERT(binary_size <= ERTS_UWORD_MAX / 8);
+
+            erts_copy_bits(binary_data, bit_offset, 1,
+                (byte*)&(state->acc)->orig_bytes[state->acc_size], 0, 1,
+                spill * 8);
+        }
 
         state->acc_size += spill;
 
diff --git a/erts/emulator/test/iovec_SUITE.erl b/erts/emulator/test/iovec_SUITE.erl
index baf1174a14..76adbe4b5e 100644
--- a/erts/emulator/test/iovec_SUITE.erl
+++ b/erts/emulator/test/iovec_SUITE.erl
@@ -25,7 +25,7 @@
 -export([integer_lists/1, binary_lists/1, empty_lists/1, empty_binary_lists/1,
          mixed_lists/1, improper_lists/1, illegal_lists/1, cons_bomb/1,
          sub_binary_lists/1, iolist_to_iovec_idempotence/1,
-         iolist_to_iovec_correctness/1]).
+         iolist_to_iovec_correctness/1, unaligned_sub_binaries/1]).
 
 -include_lib("common_test/include/ct.hrl").
 
@@ -36,7 +36,8 @@ suite() ->
 all() ->
     [integer_lists, binary_lists, empty_lists, empty_binary_lists, mixed_lists,
      sub_binary_lists, illegal_lists, improper_lists, cons_bomb,
-     iolist_to_iovec_idempotence, iolist_to_iovec_correctness].
+     iolist_to_iovec_idempotence, iolist_to_iovec_correctness,
+     unaligned_sub_binaries].
 
 init_per_suite(Config) ->
     Config.
@@ -123,6 +124,15 @@ iolist_to_iovec_correctness(Config) when is_list(Config) ->
     true = is_iolist_equal(Optimized, Variations),
     ok.
 
+unaligned_sub_binaries(Config) when is_list(Config) ->
+    UnalignedBins = [gen_unaligned_binary(I) || I <- lists:seq(32, 4 bsl 10, 512)],
+    UnalignedVariations = gen_variations(UnalignedBins),
+
+    Optimized = erlang:iolist_to_iovec(UnalignedVariations),
+
+    true = is_iolist_equal(Optimized, UnalignedVariations),
+    ok.
+
 illegality_test(Fun, Variations) ->
     [{'EXIT',{badarg, _}} = (catch Fun(Variation)) || Variation <- Variations],
     ok.
@@ -138,6 +148,13 @@ equivalence_test(Fun, [Head | _] = Variations) ->
 is_iolist_equal(A, B) ->
     iolist_to_binary(A) =:= iolist_to_binary(B).
 
+gen_unaligned_binary(Size) ->
+    Bin0 = << <<I>> || I <- lists:seq(1, Size) >>,
+    <<0:3,Bin:Size/binary,31:5>> = id(<<0:3,Bin0/binary,31:5>>),
+    Bin.
+
+id(I) -> I.
+
 %% Generates a bunch of lists whose contents will be equal to Base repeated a
 %% few times. The lists only differ by their structure, so their reduction to
 %% a simpler format should yield the same result.
-- 
cgit v1.2.3