diff options
109 files changed, 9078 insertions, 490 deletions
diff --git a/OTP_VERSION b/OTP_VERSION index a56f451a4e..06d4ac2bfd 100644 --- a/OTP_VERSION +++ b/OTP_VERSION @@ -1,2 +1 @@ 21.0-rc0 - diff --git a/bootstrap/bin/no_dot_erlang.boot b/bootstrap/bin/no_dot_erlang.boot Binary files differindex fc355b9c10..a49298d260 100644 --- a/bootstrap/bin/no_dot_erlang.boot +++ b/bootstrap/bin/no_dot_erlang.boot diff --git a/bootstrap/bin/start.boot b/bootstrap/bin/start.boot Binary files differindex fc355b9c10..a49298d260 100644 --- a/bootstrap/bin/start.boot +++ b/bootstrap/bin/start.boot diff --git a/bootstrap/bin/start_clean.boot b/bootstrap/bin/start_clean.boot Binary files differindex fc355b9c10..a49298d260 100644 --- a/bootstrap/bin/start_clean.boot +++ b/bootstrap/bin/start_clean.boot diff --git a/bootstrap/lib/compiler/ebin/compile.beam b/bootstrap/lib/compiler/ebin/compile.beam Binary files differindex 5387a0074d..f6ad883ed8 100644 --- a/bootstrap/lib/compiler/ebin/compile.beam +++ b/bootstrap/lib/compiler/ebin/compile.beam diff --git a/bootstrap/lib/stdlib/ebin/stdlib.app b/bootstrap/lib/stdlib/ebin/stdlib.app index 07309b88c4..a8abad00ea 100644 --- a/bootstrap/lib/stdlib/ebin/stdlib.app +++ b/bootstrap/lib/stdlib/ebin/stdlib.app @@ -101,6 +101,7 @@ timer, unicode, unicode_util, + uri_string, win32reg, zip]}, {registered,[timer_server,rsh_starter,take_over_monitor,pool_master, diff --git a/bootstrap/lib/stdlib/ebin/uri_string.beam b/bootstrap/lib/stdlib/ebin/uri_string.beam Binary files differnew file mode 100644 index 0000000000..f60b12037a --- /dev/null +++ b/bootstrap/lib/stdlib/ebin/uri_string.beam diff --git a/erts/doc/src/erlang.xml b/erts/doc/src/erlang.xml index d0588fe3c1..c77f426919 100644 --- a/erts/doc/src/erlang.xml +++ b/erts/doc/src/erlang.xml @@ -4730,11 +4730,11 @@ RealSystem = system + MissedSystem</code> </func> <func> - <name name="process_flag" arity="2" clause_i="3"/> + <name name="process_flag" arity="2" clause_i="3" + anchor="process_flag_min_heap_size"/> <fsummary>Set process flag min_heap_size for the calling process. </fsummary> <desc> - <marker id="process_flag_min_heap_size"/> <p>Changes the minimum heap size for the calling process.</p> <p>Returns the old value of the flag.</p> </desc> @@ -4752,12 +4752,12 @@ RealSystem = system + MissedSystem</code> </func> <func> - <name name="process_flag" arity="2" clause_i="5"/> + <name name="process_flag" arity="2" clause_i="5" + anchor="process_flag_max_heap_size"/> <fsummary>Set process flag max_heap_size for the calling process. </fsummary> <type name="max_heap_size"/> <desc> - <marker id="process_flag_max_heap_size"/> <p>This flag sets the maximum heap size for the calling process. If <c><anno>MaxHeapSize</anno></c> is an integer, the system default values for <c>kill</c> and <c>error_logger</c> are used. @@ -4826,12 +4826,12 @@ RealSystem = system + MissedSystem</code> </func> <func> - <name name="process_flag" arity="2" clause_i="6"/> + <name name="process_flag" arity="2" clause_i="6" + anchor="process_flag_message_queue_data"/> <fsummary>Set process flag message_queue_data for the calling process. </fsummary> <type name="message_queue_data"/> <desc> - <marker id="process_flag_message_queue_data"/> <p>This flag determines how messages in the message queue are stored, as follows:</p> <taglist> @@ -4868,11 +4868,12 @@ RealSystem = system + MissedSystem</code> </func> <func> - <name name="process_flag" arity="2" clause_i="7"/> + <name name="process_flag" arity="2" clause_i="7" + anchor="process_flag_priority"/> <fsummary>Set process flag priority for the calling process.</fsummary> <type name="priority_level"/> <desc> - <p><marker id="process_flag_priority"></marker> + <p> Sets the process priority. <c><anno>Level</anno></c> is an atom. Four priority levels exist: <c>low</c>, <c>normal</c>, <c>high</c>, and <c>max</c>. Default @@ -6347,10 +6348,10 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="1"/> + <name name="statistics" arity="1" clause_i="1" + anchor="statistics_active_tasks"/> <fsummary>Information about active processes and ports.</fsummary> <desc> - <marker id="statistics_active_tasks"></marker> <p>Returns the same as <seealso marker="#statistics_active_tasks_all"> <c>statistics(active_tasks_all)</c></seealso> @@ -6362,10 +6363,10 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="2"/> + <name name="statistics" arity="1" clause_i="2" + anchor="statistics_active_tasks_all"/> <fsummary>Information about active processes and ports.</fsummary> <desc> - <marker id="statistics_active_tasks_all"></marker> <p>Returns a list where each element represents the amount of active processes and ports on each run queue and its associated schedulers. That is, the number of processes and @@ -6413,10 +6414,10 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="4"/> + <name name="statistics" arity="1" clause_i="4" + anchor="statistics_exact_reductions"/> <fsummary>Information about exact reductions.</fsummary> <desc> - <marker id="statistics_exact_reductions"></marker> <p>Returns the number of exact reductions.</p> <note> <p><c>statistics(exact_reductions)</c> is @@ -6451,10 +6452,10 @@ true</pre> </func> <func> - <name name="statistics" arity="1" clause_i="7"/> + <name name="statistics" arity="1" clause_i="7" + anchor="statistics_microstate_accounting"/> <fsummary>Information about microstate accounting.</fsummary> <desc> - <marker id="statistics_microstate_accounting"></marker> <p>Microstate accounting can be used to measure how much time the Erlang runtime system spends doing various tasks. It is designed to be as lightweight as possible, but some overhead exists when this @@ -6594,10 +6595,10 @@ lists:map( </func> <func> - <name name="statistics" arity="1" clause_i="8"/> + <name name="statistics" arity="1" clause_i="8" + anchor="statistics_reductions"/> <fsummary>Information about reductions.</fsummary> <desc> - <marker id="statistics_reductions"></marker> <p>Returns information about reductions, for example:</p> <pre> > <input>statistics(reductions).</input> @@ -6613,9 +6614,10 @@ lists:map( </func> <func> - <name name="statistics" arity="1" clause_i="9"/> + <name name="statistics" arity="1" clause_i="9" + anchor="statistics_run_queue"/> <fsummary>Information about the run-queues.</fsummary> - <desc><marker id="statistics_run_queue"></marker> + <desc> <p>Returns the total length of all normal run-queues. That is, the number of processes and ports that are ready to run on all available normal run-queues. Dirty run queues are not part of the @@ -6629,9 +6631,10 @@ lists:map( </func> <func> - <name name="statistics" arity="1" clause_i="10"/> + <name name="statistics" arity="1" clause_i="10" + anchor="statistics_run_queue_lengths"/> <fsummary>Information about the run-queue lengths.</fsummary> - <desc><marker id="statistics_run_queue_lengths"></marker> + <desc> <p>Returns the same as <seealso marker="#statistics_run_queue_lengths_all"> <c>statistics(run_queue_lengths_all)</c></seealso> @@ -6643,9 +6646,10 @@ lists:map( </func> <func> - <name name="statistics" arity="1" clause_i="11"/> + <name name="statistics" arity="1" clause_i="11" + anchor="statistics_run_queue_lengths_all"/> <fsummary>Information about the run-queue lengths.</fsummary> - <desc><marker id="statistics_run_queue_lengths_all"></marker> + <desc> <p>Returns a list where each element represents the amount of processes and ports ready to run for each run queue. Values for normal run queues are located first in the @@ -6703,10 +6707,10 @@ lists:map( </func> <func> - <name name="statistics" arity="1" clause_i="13"/> + <name name="statistics" arity="1" clause_i="13" + anchor="statistics_scheduler_wall_time"/> <fsummary>Information about each schedulers work time.</fsummary> <desc> - <marker id="statistics_scheduler_wall_time"></marker> <p>Returns a list of tuples with <c>{<anno>SchedulerId</anno>, <anno>ActiveTime</anno>, <anno>TotalTime</anno>}</c>, where @@ -6824,10 +6828,10 @@ ok </func> <func> - <name name="statistics" arity="1" clause_i="14"/> + <name name="statistics" arity="1" clause_i="14" + anchor="statistics_scheduler_wall_time_all"/> <fsummary>Information about each schedulers work time.</fsummary> <desc> - <marker id="statistics_scheduler_wall_time_all"></marker> <p>The same as <seealso marker="#statistics_scheduler_wall_time"><c>statistics(scheduler_wall_time)</c></seealso>, except that it also include information about all dirty I/O @@ -6849,9 +6853,10 @@ ok </desc> </func> <func> - <name name="statistics" arity="1" clause_i="15"/> + <name name="statistics" arity="1" clause_i="15" + anchor="statistics_total_active_tasks"/> <fsummary>Information about active processes and ports.</fsummary> - <desc><marker id="statistics_total_active_tasks"></marker> + <desc> <p>The same as calling <c>lists:sum(</c><seealso marker="#statistics_active_tasks"><c>statistics(active_tasks)</c></seealso><c>)</c>, but more efficient.</p> @@ -6859,9 +6864,10 @@ ok </func> <func> - <name name="statistics" arity="1" clause_i="16"/> + <name name="statistics" arity="1" clause_i="16" + anchor="statistics_total_active_tasks_all"/> <fsummary>Information about active processes and ports.</fsummary> - <desc><marker id="statistics_total_active_tasks_all"></marker> + <desc> <p>The same as calling <c>lists:sum(</c><seealso marker="#statistics_active_tasks_all"><c>statistics(active_tasks_all)</c></seealso><c>)</c>, but more efficient.</p> @@ -6869,9 +6875,10 @@ ok </func> <func> - <name name="statistics" arity="1" clause_i="17"/> + <name name="statistics" arity="1" clause_i="17" + anchor="statistics_total_run_queue_lengths"/> <fsummary>Information about the run-queue lengths.</fsummary> - <desc><marker id="statistics_total_run_queue_lengths"></marker> + <desc> <p>The same as calling <c>lists:sum(</c><seealso marker="#statistics_run_queue_lengths"><c>statistics(run_queue_lengths)</c></seealso><c>)</c>, but more efficient.</p> @@ -6879,9 +6886,10 @@ ok </func> <func> - <name name="statistics" arity="1" clause_i="18"/> + <name name="statistics" arity="1" clause_i="18" + anchor="statistics_total_run_queue_lengths_all"/> <fsummary>Information about the run-queue lengths.</fsummary> - <desc><marker id="statistics_total_run_queue_lengths_all"></marker> + <desc> <p>The same as calling <c>lists:sum(</c><seealso marker="#statistics_run_queue_lengths_all"><c>statistics(run_queue_lengths_all)</c></seealso><c>)</c>, but more efficient.</p> @@ -7029,7 +7037,8 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="2"/> + <name name="system_flag" arity="2" clause_i="2" + anchor="system_flag_cpu_topology"/> <fsummary>Set system flag <c>cpu_topology</c>.</fsummary> <type name="cpu_topology"/> <type name="level_entry"/> @@ -7038,7 +7047,7 @@ ok <type name="info_list"/> <desc> <warning> - <p><marker id="system_flag_cpu_topology"></marker> + <p> <em>This argument is deprecated.</em> Instead of using this argument, use command-line argument <seealso marker="erts:erl#+sct"><c>+sct</c></seealso> in @@ -7076,10 +7085,11 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="3"/> + <name name="system_flag" arity="2" clause_i="3" + anchor="system_flag_dirty_cpu_schedulers_online"/> <fsummary>Set system_flag_dirty_cpu_schedulers_online.</fsummary> <desc> - <p><marker id="system_flag_dirty_cpu_schedulers_online"></marker> + <p> Sets the number of dirty CPU schedulers online. Range is <c><![CDATA[1 <= DirtyCPUSchedulersOnline <= N]]></c>, where <c>N</c> is the smallest of the return values of @@ -7123,10 +7133,11 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="5"/> + <name name="system_flag" arity="2" clause_i="5" + anchor="system_flag_microstate_accounting"/> <fsummary>Set system flag microstate_accounting.</fsummary> <desc> - <p><marker id="system_flag_microstate_accounting"></marker> + <p> Turns on/off microstate accounting measurements. When passing reset, all counters are reset to 0.</p> <p>For more information see @@ -7168,28 +7179,29 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="8"/> + <name name="system_flag" arity="2" clause_i="8" + anchor="system_flag_max_heap_size"/> <fsummary>Set system flag max_heap_size.</fsummary> <type name="max_heap_size"/> <desc> - <marker id="system_flag_max_heap_size"></marker> <p> Sets the default maximum heap size settings for processes. The size is specified in words. The new <c>max_heap_size</c> effects only processes spawned efter the change has been made. <c>max_heap_size</c> can be set for individual processes using <seealso marker="#spawn_opt/4"><c>spawn_opt/2,3,4</c></seealso> or - <seealso marker="#process_flag_message_queue_data"> + <seealso marker="#process_flag_max_heap_size"> <c>process_flag/2</c></seealso>.</p> <p>Returns the old value of the flag.</p> </desc> </func> <func> - <name name="system_flag" arity="2" clause_i="9"/> + <name name="system_flag" arity="2" clause_i="9" + anchor="system_flag_multi_scheduling"/> <fsummary>Set system flag multi_scheduling.</fsummary> <desc> - <p><marker id="system_flag_multi_scheduling"></marker> + <p> If multi-scheduling is enabled, more than one scheduler thread is used by the emulator. Multi-scheduling can be blocked in two different ways. Either all schedulers but @@ -7241,12 +7253,13 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="10"/> + <name name="system_flag" arity="2" clause_i="10" + anchor="system_flag_scheduler_bind_type"/> <fsummary>Set system flag scheduler_bind_type.</fsummary> <type name="scheduler_bind_type"/> <desc> <warning> - <p><marker id="system_flag_scheduler_bind_type"></marker> + <p> <em>This argument is deprecated.</em> Instead of using this argument, use command-line argument <seealso marker="erts:erl#+sbt"><c>+sbt</c></seealso> in @@ -7367,10 +7380,11 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="11"/> + <name name="system_flag" arity="2" clause_i="11" + anchor="system_flag_scheduler_wall_time"/> <fsummary>Set system flag scheduler_wall_time.</fsummary> <desc> - <p><marker id="system_flag_scheduler_wall_time"></marker> + <p> Turns on or off scheduler wall time measurements.</p> <p>For more information, see <seealso marker="#statistics_scheduler_wall_time"> @@ -7379,10 +7393,11 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="12"/> + <name name="system_flag" arity="2" clause_i="12" + anchor="system_flag_schedulers_online"/> <fsummary>Set system flag schedulers_online.</fsummary> <desc> - <p><marker id="system_flag_schedulers_online"></marker> + <p> Sets the number of schedulers online. Range is <c><![CDATA[1 <= SchedulersOnline <= erlang:system_info(schedulers)]]></c>.</p> @@ -7421,10 +7436,11 @@ ok </func> <func> - <name name="system_flag" arity="2" clause_i="14"/> + <name name="system_flag" arity="2" clause_i="14" + anchor="system_flag_time_offset"/> <fsummary>Finalize the time offset.</fsummary> <desc> - <p><marker id="system_flag_time_offset"></marker> + <p> Finalizes the <seealso marker="#time_offset/0">time offset</seealso> when <seealso marker="time_correction#Single_Time_Warp_Mode">single time warp mode</seealso> is used. If another time warp mode @@ -7597,7 +7613,8 @@ ok </func> <func> - <name name="system_info" arity="1" clause_i="12"/> + <name name="system_info" arity="1" clause_i="12" + anchor="system_info_cpu_topology"/> <name name="system_info" arity="1" clause_i="13"/> <fsummary>Information about the CPU topology of the system.</fsummary> <type name="cpu_topology"/> @@ -7625,7 +7642,6 @@ ok </type_desc> <desc> <marker id="system_info_cpu_topology_tags"></marker> - <marker id="system_info_cpu_topology"></marker> <p>Returns various information about the CPU topology of the current system (emulator) as specified by <c><anno>Item</anno></c>:</p> diff --git a/erts/doc/src/notes.xml b/erts/doc/src/notes.xml index 25b72ce774..bf41b61136 100644 --- a/erts/doc/src/notes.xml +++ b/erts/doc/src/notes.xml @@ -31,6 +31,49 @@ </header> <p>This document describes the changes made to the ERTS application.</p> +<section><title>Erts 9.1.4</title> + + <section><title>Fixed Bugs and Malfunctions</title> + <list> + <item> + <p>Microstate accounting sometimes produced incorrect + results for dirty schedulers.</p> + <p> + Own Id: OTP-14707</p> + </item> + <item> + <p>Fixed a regression in <c>zlib:gunzip/1</c> that + prevented it from working when the decompressed size was + a perfect multiple of 16384. This regression was + introduced in 20.1.1</p> + <p> + Own Id: OTP-14730 Aux Id: ERL-507 </p> + </item> + <item> + <p>Fixed a memory corruption bug in + <c>enif_inspect_iovec</c>; writable binaries stayed + writable after entering the iovec.</p> + <p> + Own Id: OTP-14745</p> + </item> + <item> + <p>Fixed a crash in <c>enif_inspect_iovec</c> on + encountering empty binaries.</p> + <p> + Own Id: OTP-14750</p> + </item> + <item> + <p><c>zlib:deflateParams/3</c> will no longer return + <c>buf_error</c> when called after <c>zlib:deflate/2</c> + with zlib <c>1.2.11</c>.</p> + <p> + Own Id: OTP-14751</p> + </item> + </list> + </section> + +</section> + <section><title>Erts 9.1.3</title> <section><title>Fixed Bugs and Malfunctions</title> diff --git a/erts/emulator/beam/beam_bp.c b/erts/emulator/beam/beam_bp.c index fe1e15701b..0832b3f374 100644 --- a/erts/emulator/beam/beam_bp.c +++ b/erts/emulator/beam/beam_bp.c @@ -998,7 +998,9 @@ do_call_trace(Process* c_p, ErtsCodeInfo* info, Eterm* reg, fixup_cp_before_trace(c_p, &return_to_trace); + ERTS_UNREQ_PROC_MAIN_LOCK(c_p); flags = erts_call_trace(c_p, info, ms, reg, local, &tracer); + ERTS_REQ_PROC_MAIN_LOCK(c_p); /* restore cp after potential fixup */ c_p->cp = cp_save; diff --git a/erts/emulator/beam/beam_emu.c b/erts/emulator/beam/beam_emu.c index 60d0008d8f..bc95ccec52 100644 --- a/erts/emulator/beam/beam_emu.c +++ b/erts/emulator/beam/beam_emu.c @@ -400,12 +400,13 @@ static BeamInstr* apply_fun(Process* p, Eterm fun, Eterm args, Eterm* reg) NOINLINE; static Eterm new_fun(Process* p, Eterm* reg, ErlFunEntry* fe, int num_free) NOINLINE; -static Eterm new_map(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* ptr) NOINLINE; -static Eterm new_small_map_lit(Process* p, Eterm* reg, Eterm keys_literal, +static Eterm erts_gc_new_map(Process* p, Eterm* reg, Uint live, + Uint n, BeamInstr* ptr) NOINLINE; +static Eterm erts_gc_new_small_map_lit(Process* p, Eterm* reg, Eterm keys_literal, Uint live, BeamInstr* ptr) NOINLINE; -static Eterm update_map_assoc(Process* p, Eterm* reg, Uint live, +static Eterm erts_gc_update_map_assoc(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* new_p) NOINLINE; -static Eterm update_map_exact(Process* p, Eterm* reg, Uint live, +static Eterm erts_gc_update_map_exact(Process* p, Eterm* reg, Uint live, Uint n, Eterm* new_p) NOINLINE; static Eterm get_map_element(Eterm map, Eterm key); static Eterm get_map_element_hash(Eterm map, Eterm key, Uint32 hx); @@ -2755,7 +2756,7 @@ do { \ static Eterm -new_map(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* ptr) +erts_gc_new_map(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* ptr) { Uint i; Uint need = n + 1 /* hdr */ + 1 /*size*/ + 1 /* ptr */ + 1 /* arity */; @@ -2812,7 +2813,8 @@ new_map(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* ptr) } static Eterm -new_small_map_lit(Process* p, Eterm* reg, Eterm keys_literal, Uint live, BeamInstr* ptr) +erts_gc_new_small_map_lit(Process* p, Eterm* reg, Eterm keys_literal, + Uint live, BeamInstr* ptr) { Eterm* keys = tuple_val(keys_literal); Uint n = arityval(*keys); @@ -2846,7 +2848,8 @@ new_small_map_lit(Process* p, Eterm* reg, Eterm keys_literal, Uint live, BeamIns } static Eterm -update_map_assoc(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* new_p) +erts_gc_update_map_assoc(Process* p, Eterm* reg, Uint live, + Uint n, BeamInstr* new_p) { Uint num_old; Uint num_updates; @@ -2892,7 +2895,7 @@ update_map_assoc(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* new_p) */ if (num_old == 0) { - return new_map(p, reg, live, n, new_p); + return erts_gc_new_map(p, reg, live, n, new_p); } /* @@ -3048,7 +3051,7 @@ update_map_assoc(Process* p, Eterm* reg, Uint live, Uint n, BeamInstr* new_p) */ static Eterm -update_map_exact(Process* p, Eterm* reg, Uint live, Uint n, Eterm* new_p) +erts_gc_update_map_exact(Process* p, Eterm* reg, Uint live, Uint n, Eterm* new_p) { Uint i; Uint num_old; diff --git a/erts/emulator/beam/beam_load.c b/erts/emulator/beam/beam_load.c index 7331c331a6..4fcfea527c 100644 --- a/erts/emulator/beam/beam_load.c +++ b/erts/emulator/beam/beam_load.c @@ -2384,8 +2384,18 @@ load_code(LoaderState* stp) code[ci++] = NIL; break; case TAG_q: - new_literal_patch(stp, ci); - code[ci++] = tmp_op->a[arg].val; + { + BeamInstr val = tmp_op->a[arg].val; + Eterm term = stp->literals[val].term; + new_literal_patch(stp, ci); + code[ci++] = val; + switch (loader_tag(term)) { + case LOADER_X_REG: + case LOADER_Y_REG: + LoadError1(stp, "the term '%T' would be confused " + "with a register", term); + } + } break; default: LoadError1(stp, "bad tag %d for general source", diff --git a/erts/emulator/beam/erl_nif.c b/erts/emulator/beam/erl_nif.c index f7f12efe28..c21b139cfa 100644 --- a/erts/emulator/beam/erl_nif.c +++ b/erts/emulator/beam/erl_nif.c @@ -3323,36 +3323,38 @@ static int examine_iovec_term(Eterm list, UWord max_length, iovec_slice_t *resul size = binary_size(binary); binary_header = binary_val(binary); - /* If we're a sub-binary we'll need to check our underlying binary to - * determine whether we're on-heap or not. */ - if(thing_subtag(*binary_header) == SUB_BINARY_SUBTAG) { - ErlSubBin *sb = (ErlSubBin*)binary_header; - - /* Reject bitstrings */ - if((sb->bitoffs + sb->bitsize) > 0) { - return 0; + if (size > 0) { + /* If we're a sub-binary we'll need to check our underlying binary + * to determine whether we're on-heap or not. */ + if (thing_subtag(*binary_header) == SUB_BINARY_SUBTAG) { + ErlSubBin *sb = (ErlSubBin*)binary_header; + + /* Reject bitstrings */ + if((sb->bitoffs + sb->bitsize) > 0) { + return 0; + } + + ASSERT(size <= binary_size(sb->orig)); + binary_header = binary_val(sb->orig); } - ASSERT(size <= binary_size(sb->orig)); - binary_header = binary_val(sb->orig); - } - - if(thing_subtag(*binary_header) == HEAP_BINARY_SUBTAG) { - ASSERT(size <= ERL_ONHEAP_BIN_LIMIT); + if (thing_subtag(*binary_header) == HEAP_BINARY_SUBTAG) { + ASSERT(size <= ERL_ONHEAP_BIN_LIMIT); - result->iovec_len += 1; - result->onheap_size += size; - } else { - ASSERT(thing_subtag(*binary_header) == REFC_BINARY_SUBTAG); + result->iovec_len += 1; + result->onheap_size += size; + } else { + ASSERT(thing_subtag(*binary_header) == REFC_BINARY_SUBTAG); - result->iovec_len += 1 + size / MAX_SYSIOVEC_IOVLEN; - result->offheap_size += size; + result->iovec_len += 1 + size / MAX_SYSIOVEC_IOVLEN; + result->offheap_size += size; + } } result->sublist_length += 1; lookahead = CDR(cell); - if(result->sublist_length >= max_length) { + if (result->sublist_length >= max_length) { break; } } @@ -3385,6 +3387,10 @@ static void inspect_raw_binary_data(Eterm binary, ErlNifBinary *result) { if (thing_subtag(*parent_header) == REFC_BINARY_SUBTAG) { ProcBin *pb = (ProcBin*)parent_header; + if (pb->flags & (PB_IS_WRITABLE | PB_ACTIVE_WRITER)) { + erts_emasculate_writable_binary(pb); + } + ASSERT(pb->val != NULL); ASSERT(byte_offset < pb->size); ASSERT(&pb->bytes[byte_offset] >= (byte*)(pb->val)->orig_bytes); @@ -3428,7 +3434,7 @@ static int fill_iovec_with_slice(ErlNifEnv *env, /* If this isn't a refc binary, copy its contents to the onheap buffer * and reference that instead. */ - if (raw_data.ref_bin == NULL) { + if (raw_data.size > 0 && raw_data.ref_bin == NULL) { ASSERT(onheap_offset < onheap_data.size); ASSERT(slice->onheap_size > 0); @@ -3439,12 +3445,11 @@ static int fill_iovec_with_slice(ErlNifEnv *env, raw_data.ref_bin = onheap_data.ref_bin; } - ASSERT(raw_data.ref_bin != NULL); - while (raw_data.size > 0) { UWord chunk_len = MIN(raw_data.size, MAX_SYSIOVEC_IOVLEN); ASSERT(iovec_idx < iovec->iovcnt); + ASSERT(raw_data.ref_bin != NULL); iovec->iov[iovec_idx].iov_base = raw_data.data; iovec->iov[iovec_idx].iov_len = chunk_len; diff --git a/erts/emulator/beam/erl_process.c b/erts/emulator/beam/erl_process.c index 6654468fb6..3c0a126fe2 100644 --- a/erts/emulator/beam/erl_process.c +++ b/erts/emulator/beam/erl_process.c @@ -10824,7 +10824,7 @@ request_system_task(Process *c_p, Eterm requester, Eterm target, goto badarg; req_type = tp[1]; req_id = tp[2]; - req_id_sz = is_immed(req_id) ? req_id : size_object(req_id); + req_id_sz = is_immed(req_id) ? 0 : size_object(req_id); tot_sz = req_id_sz; for (i = 0; i < ERTS_MAX_PROC_SYS_TASK_ARGS; i++) { int tix = 3 + i; diff --git a/erts/emulator/beam/macros.tab b/erts/emulator/beam/macros.tab index e0b5f56b53..494fe8961e 100644 --- a/erts/emulator/beam/macros.tab +++ b/erts/emulator/beam/macros.tab @@ -20,13 +20,12 @@ // // -// Use if there is a garbage collection before storing to a -// general destination (either X or Y register). +// Define a regular expression that will match instructions that +// perform GC. That will allow beam_makeops to check for instructions +// that don't use $REFRESH_GEN_DEST() when they should. // -REFRESH_GEN_DEST() { - dst_ptr = REG_TARGET_PTR(dst); -} +GC_REGEXP=erts_garbage_collect|erts_gc|GcBifFunction; // $Offset is relative to the start of the instruction (not to the // location of the failure label reference). Since combined diff --git a/erts/emulator/beam/map_instrs.tab b/erts/emulator/beam/map_instrs.tab index bbb2f49b66..c594a87298 100644 --- a/erts/emulator/beam/map_instrs.tab +++ b/erts/emulator/beam/map_instrs.tab @@ -31,7 +31,7 @@ new_map(Dst, Live, N) { Eterm res; HEAVY_SWAPOUT; - res = new_map(c_p, reg, $Live, $N, $NEXT_INSTRUCTION); + res = erts_gc_new_map(c_p, reg, $Live, $N, $NEXT_INSTRUCTION); HEAVY_SWAPIN; $REFRESH_GEN_DEST(); $Dst = res; @@ -44,7 +44,7 @@ i_new_small_map_lit(Dst, Live, Keys) { Eterm keys = $Keys; HEAVY_SWAPOUT; - res = new_small_map_lit(c_p, reg, keys, $Live, $NEXT_INSTRUCTION); + res = erts_gc_new_small_map_lit(c_p, reg, keys, $Live, $NEXT_INSTRUCTION); HEAVY_SWAPIN; $REFRESH_GEN_DEST(); $Dst = res; @@ -133,7 +133,7 @@ update_map_assoc(Src, Dst, Live, N) { reg[live] = $Src; HEAVY_SWAPOUT; - res = update_map_assoc(c_p, reg, live, $N, $NEXT_INSTRUCTION); + res = erts_gc_update_map_assoc(c_p, reg, live, $N, $NEXT_INSTRUCTION); HEAVY_SWAPIN; ASSERT(is_value(res)); $REFRESH_GEN_DEST(); @@ -147,7 +147,7 @@ update_map_exact(Fail, Src, Dst, Live, N) { reg[live] = $Src; HEAVY_SWAPOUT; - res = update_map_exact(c_p, reg, live, $N, $NEXT_INSTRUCTION); + res = erts_gc_update_map_exact(c_p, reg, live, $N, $NEXT_INSTRUCTION); HEAVY_SWAPIN; if (is_value(res)) { $REFRESH_GEN_DEST(); diff --git a/erts/emulator/beam/ops.tab b/erts/emulator/beam/ops.tab index 7a2c39b3a8..a560bde920 100644 --- a/erts/emulator/beam/ops.tab +++ b/erts/emulator/beam/ops.tab @@ -511,8 +511,6 @@ put_list y x x put_list y y x put_list x y x -put_list y x x - # put_list SrcReg Constant Dst put_list x c x diff --git a/erts/emulator/hipe/hipe_amd64_bifs.m4 b/erts/emulator/hipe/hipe_amd64_bifs.m4 index 6d998c4b55..cf4c59c9af 100644 --- a/erts/emulator/hipe/hipe_amd64_bifs.m4 +++ b/erts/emulator/hipe/hipe_amd64_bifs.m4 @@ -462,42 +462,6 @@ ASYM($1): TYPE_FUNCTION(ASYM($1)) #endif') -/* - * nogc_bif_interface_1(nbif_name, cbif_name) - * - * Generate native interface for a bif with implicit P - * The bif can fail but cannot do GC. - */ - -define(nogc_bif_interface_1, -` -#ifndef HAVE_$1 -#`define' HAVE_$1 - TEXT - .align 4 - GLOBAL(ASYM($1)) -ASYM($1): - /* set up the parameters */ - movq P, %rdi - NBIF_ARG(%rsi,1,0) - - /* make the call on the C stack */ - SWITCH_ERLANG_TO_C - pushq %rsi - movq %rsp, %rsi /* Eterm* BIF__ARGS */ - sub $(8), %rsp /* stack frame 16-byte alignment */ - CALL_BIF($2) - add $(1*8 + 8), %rsp - SWITCH_C_TO_ERLANG - - /* throw exception if failure, otherwise return */ - TEST_GOT_EXN - jz nbif_1_simple_exception - NBIF_RET(1) - SET_SIZE(ASYM($1)) - TYPE_FUNCTION(ASYM($1)) -#endif') - /* * noproc_primop_interface_0(nbif_name, cbif_name) diff --git a/erts/emulator/hipe/hipe_bif0.tab b/erts/emulator/hipe/hipe_bif0.tab index 4f73770d24..0380e8c795 100644 --- a/erts/emulator/hipe/hipe_bif0.tab +++ b/erts/emulator/hipe/hipe_bif0.tab @@ -135,7 +135,6 @@ atom bs_utf16_size atom bs_put_utf16be atom bs_put_utf16le atom bs_get_utf16 -atom bs_validate_unicode atom bs_validate_unicode_retract atom emulate_fpe atom emasculate_binary diff --git a/erts/emulator/hipe/hipe_bif_list.m4 b/erts/emulator/hipe/hipe_bif_list.m4 index 23b6709cd0..625d8486fd 100644 --- a/erts/emulator/hipe/hipe_bif_list.m4 +++ b/erts/emulator/hipe/hipe_bif_list.m4 @@ -248,11 +248,6 @@ nofail_primop_interface_3(nbif_bs_get_float_2, erts_bs_get_float_2) standard_bif_interface_3(nbif_bs_put_utf8, hipe_bs_put_utf8) standard_bif_interface_3(nbif_bs_put_utf16be, hipe_bs_put_utf16be) standard_bif_interface_3(nbif_bs_put_utf16le, hipe_bs_put_utf16le) -ifdef(`nogc_bif_interface_1',` -nogc_bif_interface_1(nbif_bs_validate_unicode, hipe_bs_validate_unicode) -',` -standard_bif_interface_1(nbif_bs_validate_unicode, hipe_bs_validate_unicode) -') /* * Bit-syntax primops without any P parameter. diff --git a/erts/emulator/hipe/hipe_debug.c b/erts/emulator/hipe/hipe_debug.c index 222a11db3d..cfe60b379e 100644 --- a/erts/emulator/hipe/hipe_debug.c +++ b/erts/emulator/hipe/hipe_debug.c @@ -135,7 +135,9 @@ static void print_heap(Eterm *pos, Eterm *end) printf("From: 0x%0*lx to 0x%0*lx\n\r", 2*(int)sizeof(long), (unsigned long)pos, 2*(int)sizeof(long), (unsigned long)end); - printf(" | H E A P |\r\n"); + printf(" | %*s H E A P %*s |\r\n", + 2*(int)sizeof(long)-1, "", + 2*(int)sizeof(long)-1, ""); printf(" | %*s | %*s |\r\n", 2+2*(int)sizeof(long), "Address", 2+2*(int)sizeof(long), "Contents"); @@ -158,8 +160,10 @@ static void print_heap(Eterm *pos, Eterm *end) ++pos; --ari; } - } else + } else { + fflush(stdout); erts_printf("%.30T", val); + } printf("\r\n"); } printf(" |%s|%s|\r\n", dashes, dashes); diff --git a/erts/emulator/hipe/hipe_native_bif.c b/erts/emulator/hipe/hipe_native_bif.c index e444e3dc5d..99c34532b9 100644 --- a/erts/emulator/hipe/hipe_native_bif.c +++ b/erts/emulator/hipe/hipe_native_bif.c @@ -482,15 +482,6 @@ static int validate_unicode(Eterm arg) return 1; } -BIF_RETTYPE nbif_impl_hipe_bs_validate_unicode(NBIF_ALIST_1) -{ - Process *p = BIF_P; - Eterm arg = BIF_ARG_1; - if (!validate_unicode(arg)) - BIF_ERROR(p, BADARG); - return NIL; -} - Uint hipe_is_unicode(Eterm arg) { return (Uint) validate_unicode(arg); diff --git a/erts/emulator/hipe/hipe_native_bif.h b/erts/emulator/hipe/hipe_native_bif.h index 71f63875a4..d5081b8438 100644 --- a/erts/emulator/hipe/hipe_native_bif.h +++ b/erts/emulator/hipe/hipe_native_bif.h @@ -66,7 +66,6 @@ AEXTERN(Eterm,nbif_bs_utf16_size,(Eterm)); AEXTERN(Eterm,nbif_bs_put_utf16be,(Process*,Eterm,byte*,unsigned int)); AEXTERN(Eterm,nbif_bs_put_utf16le,(Process*,Eterm,byte*,unsigned int)); AEXTERN(Eterm,nbif_bs_get_utf16,(void)); -AEXTERN(Eterm,nbif_bs_validate_unicode,(Process*,Eterm)); AEXTERN(Uint,nbif_is_unicode,(Eterm)); AEXTERN(Eterm,nbif_bs_validate_unicode_retract,(void)); AEXTERN(Uint,nbif_is_divisible,(Uint,Uint)); @@ -92,7 +91,6 @@ BIF_RETTYPE nbif_impl_hipe_bs_put_utf8(NBIF_ALIST_3); Eterm hipe_bs_utf16_size(Eterm); BIF_RETTYPE nbif_impl_hipe_bs_put_utf16be(NBIF_ALIST_3); BIF_RETTYPE nbif_impl_hipe_bs_put_utf16le(NBIF_ALIST_3); -BIF_RETTYPE nbif_impl_hipe_bs_validate_unicode(NBIF_ALIST_1); Uint hipe_is_unicode(Eterm); struct erl_bin_match_buffer; int hipe_bs_validate_unicode_retract(struct erl_bin_match_buffer*, Eterm); diff --git a/erts/emulator/hipe/hipe_primops.h b/erts/emulator/hipe/hipe_primops.h index 49e6bdb026..a6abd3e011 100644 --- a/erts/emulator/hipe/hipe_primops.h +++ b/erts/emulator/hipe/hipe_primops.h @@ -63,7 +63,6 @@ PRIMOP_LIST(am_bs_utf16_size, &nbif_bs_utf16_size) PRIMOP_LIST(am_bs_put_utf16be, &nbif_bs_put_utf16be) PRIMOP_LIST(am_bs_put_utf16le, &nbif_bs_put_utf16le) PRIMOP_LIST(am_bs_get_utf16, &nbif_bs_get_utf16) -PRIMOP_LIST(am_bs_validate_unicode, &nbif_bs_validate_unicode) PRIMOP_LIST(am_is_unicode, &nbif_is_unicode) PRIMOP_LIST(am_bs_validate_unicode_retract, &nbif_bs_validate_unicode_retract) diff --git a/erts/emulator/internal_doc/beam_makeops.md b/erts/emulator/internal_doc/beam_makeops.md new file mode 100644 index 0000000000..1da8d2ab05 --- /dev/null +++ b/erts/emulator/internal_doc/beam_makeops.md @@ -0,0 +1,1846 @@ +The beam\_makeops script +======================= + +This document describes the **beam\_makeops** script. + +Introduction +------------ + +The **beam\_makeops** Perl script is used at build-time by both the +compiler and runtime system. Given a number of input files (all with +the extension `.tab`), it will generate source files used by the +Erlang compiler and by the runtime system to load and execute BEAM +instructions. + +Essentially those `.tab` files define: + +* External generic BEAM instructions. They are the instructions that +are known to both the compiler and the runtime system. Generic +instructions are stable between releases. New generic instructions +with high numbers than previous instructions can be added in major +releases. The OTP 20 release has 159 external generic instructions. + +* Internal generic instructions. They are known only to the runtime +system and can be changed at any time without compatibility issues. +They are created by transformation rules (described next). + +* Rules for transforming one or more generic instructions to other +generic instructions. The transformation rules allow combining, +splitting, and removal of instructions, as well as shuffling operands. +Because of the transformation rules, the runtime can have many +internal generic instructions that are only known to runtime system. + +* Specific BEAM instructions. The specific instructions are the +instructions that are actually executed by the runtime system. They +can be changed at any time without causing compatibility issues. +The loader translates generic instructions to specific instructions. +In general, for each generic instruction, there exists a family of +specific instructions. The OTP 20 release has 389 specific +instructions. + +* The implementation of specific instructions. + +Generic instructions have typed operands. Here are a few examples of +operands for `move/2`: + + {move,{atom,id},{x,5}}. + {move,{x,3},{x,0}}. + {move,{x,2},{y,1}}. + +When those instructions are loaded, the loader rewrites them +to specific instructions: + + move_cx id 5 + move_xx 3 0 + move_xy 2 1 + +Corresponding to each generic instruction, there is a family of +specific instructions. The types that an instance of a specific +instruction can handle are encoded in the instruction names. For +example, `move_xy` takes an X register number as the first operand and +a Y register number as the second operand. `move_cx` takes a tagged +Erlang term as the first operand and an X register number as the +second operand. + +An example: the move instruction +-------------------------------- + +Using the `move` instruction as an example, we will give a quick +tour to show the main features of **beam\_makeops**. + +In the `compiler` application, in the file `genop.tab`, there is the +following line: + + 64: move/2 + +This is a definition of an external generic BEAM instruction. Most +importantly it specifices that the opcode is 64. It also defines that +it has two operands. The BEAM assembler will use the opcode when +creating `.beam` files. The compiler does not really need the arity, +but it will use it as an internal sanity check when assembling the +BEAM code. + +Let's have a look at `ops.tab` in `erts/emulator/beam`, where the +specific `move` instructions are defined. Here are a few of them: + + move x x + move x y + move c x + +Each specific instructions is defined by following the name of the +instruction with the types for each operand. An operand type is a +single letter. For example, `x` means an X register, `y` +means a Y register, and `c` is a "constant" (a tagged term such as +an integer, an atom, or a literal). + +Now let's look at the implementation of the `move` instruction. There +are multiple files containing implementations of instructions in the +`erts/emulator/beam` directory. The `move` instruction is defined in +`instrs.tab`. It looks like this: + + move(Src, Dst) { + $Dst = $Src; + } + +The implementation for an instruction largely follows the C syntax, +except that the variables in the function head don't have any types. +The `$` before an identifier denotes a macro expansion. Thus, +`$Src` will expand to the code to pick up the source operand for +the instruction and `$Dst` to the code for the destination register. + +We will look at the code for each specific instruction in turn. To +make the code easier to understand, let's first look at the memory +layout for the instruction `{move,{atom,id},{x,5}}`: + + +--------------------+--------------------+ + I -> | 40 | &&lb_move_cx | + +--------------------+--------------------+ + | Tagged atom 'id' | + +--------------------+--------------------+ + +This example and all other examples in the document assumes a 64-bit +archictecture, and furthermore that pointers to C code fit in 32 bits. + +`I` in the BEAM virtual machine is the instruction pointer. When BEAM +executes an instruction, `I` points to the first word of the +instruction. + +`&&lb_move_cx` is the address to C code that implements `move_cx`. It +is stored in the lower 32 bits of the word. In the upper 32 bits is +the byte offset to the X register; the register number 5 has been +multiplied by the word size size 8. + +In the next word the tagged atom `id` is stored. + +With that background, we can look at the generated code for `move_cx` +in `beam_hot.h`: + + OpCase(move_cx): + { + BeamInstr next_pf = BeamCodeAddr(I[2]); + xb(BeamExtraData(I[0])) = I[1]; + I += 2; + ASSERT(VALID_INSTR(next_pf)); + GotoPF(next_pf); + } + +We will go through each line in turn. + +* `OpCase(move_cx):` defines a label for the instruction. The +`OpCase()` macro is defined in `beam_emu.c`. It will expand this line +to `lb_move_cx:`. + +* `BeamInstr next_pf = BeamCodeAddr(I[2]);` fetches the pointer to +code for the next instruction to be executed. The `BeamCodeAddr()` +macro extracts the pointer from the lower 32 bits of the instruction +word. + +* `xb(BeamExtraData(I[0])) = I[1];` is the expansion of `$Dst = $Src`. +`BeamExtraData()` is a macro that will extract the upper 32 bits from +the instruction word. In this example, it will return 40 which is the +byte offset for X register 5. The `xb()` macro will cast a byte +pointer to an `Eterm` pointer and dereference it. The `I[1]` on +the right side of the `=` fetches an Erlang term (the atom `id` in +this case). + +* `I += 2` advances the instruction pointer to the next +instruction. + +* In a debug-compiled emulator, `ASSERT(VALID_INSTR(next_pf));` makes +sure that `next_pf` is a valid instruction (that is, that it points +within the `process_main()` function in `beam_emu.c`). + +* `GotoPF(next_pf);` transfers control to the next instruction. + +Now let's look at the implementation of `move_xx`: + + OpCase(move_xx): + { + Eterm tmp_packed1 = BeamExtraData(I[0]); + BeamInstr next_pf = BeamCodeAddr(I[1]); + xb((tmp_packed1>>BEAM_TIGHT_SHIFT)) = xb(tmp_packed1&BEAM_TIGHT_MASK); + I += 1; + ASSERT(VALID_INSTR(next_pf)); + GotoPF(next_pf); + } + +We will go through the lines that are new or have changed compared to +`move_cx`. + +* `Eterm tmp_packed1 = BeamExtraData(I[0]);` picks up both X register +numbers packed into the upper 32 bits of the instruction word. + +* `BeamInstr next_pf = BeamCodeAddr(I[1]);` pre-fetches the address of +the next instruction. Note that because both X registers operands fits +into the instruction word, the next instruction is in the very next +word. + +* `xb((tmp_packed1>>BEAM_TIGHT_SHIFT)) = xb(tmp_packed1&BEAM_TIGHT_MASK);` +copies the source to the destination. (For a 64-bit architecture, +`BEAM_TIGHT_SHIFT` is 16 and `BEAM_TIGHT_MASK` is `0xFFFF`.) + +* `I += 1;` advances the instruction pointer to the next instruction. + +`move_xy` is almost identical to `move_xx`. The only difference is +the use of the `yb()` macro instead of `xb()` to reference the +destination register: + + OpCase(move_xy): + { + Eterm tmp_packed1 = BeamExtraData(I[0]); + BeamInstr next_pf = BeamCodeAddr(I[1]); + yb((tmp_packed1>>BEAM_TIGHT_SHIFT)) = xb(tmp_packed1&BEAM_TIGHT_MASK); + I += 1; + ASSERT(VALID_INSTR(next_pf)); + GotoPF(next_pf); + } + +### Transformation rules ### + +Next let's look at how we can do some optimizations using transformation +rules. For simple instructions such as `move/2`, the instruction dispatch +overhead can be substantial. A simple optimization is to combine common +instructions sequences to a single instruction. One such common sequence +is multiple `move` instructions moving X registers to Y registers. + +Using the following rule we can combine two `move` instructions +to a `move2` instruction: + + move X1=x Y1=y | move X2=x Y2=y => move2 X1 Y1 X2 Y2 + +The left side of the arrow (`=>`) is a pattern. If the pattern +matches, the matching instructions will be replaced by the +instructions on the right side. Variables in a pattern must start +with an uppercase letter just as in Erlang. A pattern variable may be +followed `=` and one or more type letters to constrain the match to +one of those types. The variables that are bound on the left side can +be used on the right side. + +We will also need to define a specific instruction and an implementation: + + # In ops.tab + move2 x y x y + + // In instrs.tab + move2(S1, D1, S2, D2) { + Eterm V1, V2; + V1 = $S1; + V2 = $S2; + $D1 = V1; + $D2 = V2; + } + +When the loader has found a match and replaced the matched instructions, +it will match the new instructions against the transformation rules. +Because of that, we can define the rule for a `move3/6` instruction +as follows: + + move2 X1=x Y1=y X2=x Y2=y | move X3=x Y3=y => \ + move3 X1 Y1 X2 Y2 X3 Y3 + +(A `\` before a newline can be used to break a long line for readability.) + +It would also be possible to define it like this: + + move X1=x Y1=y | move X2=x Y2=y | move X3=x Y3=y => \ + move3 X1 Y1 X2 Y2 X3 Y3 + +but in that case it must be defined before the rule for `move2/4` +because the first matching rule will be applied. + +One must be careful not to create infinite loops. For example, if we +for some reason would want to reverse the operand order for the `move` +instruction, we must not do like this: + + move Src Dst => move Dst Src + +The loader would swap the operands forever. To avoid the loop, we must +rename the instruction. For example: + + move Src Dst => assign Dst Src + +This concludes the quick tour of the features of **beam\_makeops**. + +Short overview of instruction loading +------------------------------------- + +To give some background to the rest of this document, here follows a +quick overview of how instructions are loaded. + +* The loader reads and decodes one instruction at a time from the BEAM +code and creates a generic instruction. Many transformation rules +must look at multiple instructions, so the loader will +keep multiple generic instructions in a linked list. + +* The loader tries to apply transformation rules against the +generic instructions in the linked list. If a rule matches, the +matched instructions will be removed and replaced with new +generic instructions constructed from the right side of the +transformation. + +* If a transformation rule matched, the loader applies the +transformation rules again. + +* If no transformation rule match, the loader will begin rewriting +the first of generic instructions to a specific instruction. + +* First the loader will search for a specific operation where the +types for all operands match the type for the generic instruction. +The first matching instruction will be selected. **beam\_makeops** +has ordered the specific instructions so that instructions with more +specific operands comes before instructions with less specific +operands. For example, `move_nx` is more specific than `move_cx`. If +the first operand is `[]` (NIL), `move_nx` will be selected. + +* Given the opcode for the selected specific instruction, the loader +looks up the pointer to the C code for the instruction and stores +in the code area for the module being loaded. + +* The loader translates each operand to a machine word and stores it +in the code area. The operand type for the selected specific +instruction guides the translation. For example, if the type is `e`, +the value of the operand is an index into an arry of external +functions and will be translated to a pointer to the export entry for +the function to call. If the type is `x`, the number of the X +register will be multiplied by the word size to produce a byte offset. + +* The loader runs the packing engine to pack multiple operands into a +single word. The packing engine is controlled by a small program, +which is a string where each character is an instruction. For +example, the code to pack the operands for `move_xy` is `"22#"` (on a +64-bit machine). That program will pack the byte offsets for both +registers into the same word as the pointer to C code. + +Running beam_makeops +-------------------- + +**beam\_makeops** is found in `$ERL_TOP/erts/emulator/utils`. Options +start with a hyphen (`-`). The options are followed by the name of +the input files. By convention, all input files have the extension +`.tab`, but is not enforced by **beam\_makeops**. + +### The -outdir option ### + +The option `-outdir Directory` specifies the output directory for +the generated files. Default is the current working directory. + +### Running beam_makeops for the compiler ### + +Give the option `-compiler` to produce output files for the compiler. +The following files will be written to the output directory: + +* `beam_opcodes.erl` - Used primarily by `beam_asm` and `beam_diasm`. + +* `beam_opcode.hrl` - Used by `beam_asm`. It contains tag definitions +used for encoding instruction operands. + +The input file should only contain the definition of BEAM_FORMAT_NUMBER +and external generic instructions. (Everything else would be ignored.) + +### Running beam_makeops for the emulator ### + +Give the option `-emulator` to produce output files for the emulator. +The following output files will be generated in the output directory. + +* `beam_hot.h`, `beam_warm.h`, `beam_cold.`h - Implementation of +instructions. Included inside the `process_main()` function in +`beam_emu.c`. + +* `beam_opcodes.c` - Defines static data used by the loader +(`beam_load.c`). Data about generic instructions, specific +instructions (including how to pack their operands), and +transformation rules are all part of this file. + +* `beam_opcodes.h` - Miscellanous preprocessor definitions, mainly +used by `beam_load.c` but also by `beam_{hot,warm,cold}.h`. + +* `beam_pred_funcs.h` - Included by `beam_load.c`. Contains defines +needed to call guard constraints in transformation rules. + +* `beam_tr_funcs.h` - Included by `beam_load.c`. Contains defines +needed to call a C function to the right of a transformation rule. + +The following options can be given: + +* `wordsize 32|64` - Defines the word size. Default is 32. + +* `code-model Model` - The code model as given to `-mcmodel` option +for GCC. Default is `unknown`. If the code model is `small` (and +the word size is 64 bits), **beam\_makeops** will pack operands +into the upper 32 bits of the instruction word. + +* `DSymbol=0|1` - Defines the value for a symbol. The symbol can be +used in `%if` and `%unless` directives. + +Syntax of .tab files +-------------------- + +### Comments ### + +Any line starting with `#` is a comment and is ignored. + +A line with `//` is also a comment. It is recommended to only +use this style of comments in files that define implementations of +instructions. + +A long line can be broken into shorter lines by a placing a`\` before +the newline. + +### Variable definitions ### + +A variable definition binds a variable to a Perl variable. It is only +meaningful to add a new definition if **beam\_makeops** is updated +at the same time to use the variable. A variable definition looks this: + +*name*=*value*[;] + +where *name* is the name of a Perl variable in **beam\_makeops**, +and *value* is the value to be given to the variable. The line +can optionally end with a `;` (to avoid messing up the +C indentation mode in Emacs). + +Here follows a description of the variables that are defined. + +#### BEAM\_FORMAT\_NUMBER #### + +`genop.tab` has the following definition: + + BEAM_FORMAT_NUMBER=0 + +It defines the version of the instruction set (which will be +included in the code header in the BEAM code). Theoretically, +the version could be bumped, and all instructions changed. +In practice, we would have two support two instruction sets +in the runtime system for at least two releases, so it will +probably never happen in practice. + +#### GC\_REGEXP #### + +In `macros.tab`, there is a definition of `GC_REGEXP`. +It will be described in [a later section](#the-gc_regexp-definition). + +### Directives ### + +There are directives to classify specific instructions depending +on how frequently used they are: + +* `%hot` - Implementation will be placed in `beam_hot.h`. Frequently +executed instructions. + +* `%warm` - Implementation will be placed in `beam_warm.h`. Binary +syntax instructions. + +* `%cold` - Implementation will be placed in `beam_cold.h`. Trace +instructions and infrequently used instructions. + +Default is `%hot`. The directives will be applied to declarations +of the specific instruction that follow. Here is an example: + + %cold + is_number f? xy + %hot + +#### Conditional compilation directives #### + +The `%if` directive includes a range of lines if a condition is +true. For example: + + %if ARCH_64 + i_bs_get_integer_32 x f? x + %endif + +The specific instruction `i_bs_get_integer_32` will only be defined +on a 64-bit machine. + +The condition can be inverted by using `%unless` instead of `%if`: + + %unless NO_FPE_SIGNALS + fcheckerror p => i_fcheckerror + i_fcheckerror + fclearerror + %endif + +It is also possible to add an `%else` clause: + + %if ARCH_64 + BS_SAFE_MUL(A, B, Fail, Dst) { + Uint64 res = ($A) * ($B); + if (res / $B != $A) { + $Fail; + } + $Dst = res; + } + %else + BS_SAFE_MUL(A, B, Fail, Dst) { + Uint64 res = (Uint64)($A) * (Uint64)($B); + if ((res >> (8*sizeof(Uint))) != 0) { + $Fail; + } + $Dst = res; + } + %endif + +#### Symbols that are defined in directives #### + +The following symbols are always defined. + +* `ARCH_64` - is 1 for a 64-bit machine, and 0 otherwise. +* `ARCH_32` - is 1 for 32-bit machine, and 1 otherwise. + +The `Makefile` for building the emulator currently defines the +following symbols by using the `-D` option on the command line for +**beam\_makeops**. + +* `NO_FPE_SIGNALS` - 1 if FPE signals are not enable in runtime system, +0 otherwise. +* `USE_VM_PROBES` - 1 if the runtime system is compiled to use VM probes (support for dtrace or systemtap), 0 otherwise. + +### Defining external generic instructions ### + +External generic BEAM instructions are known to both the compiler and +the runtime system. They remain stable between releases. A new major +release may add more external generic instructions, but must not change +the semantics for a previously defined instruction. + +The syntax for an external generic instruction is as follows: + +*opcode*: [-]*name*/*arity* + +*opcode* is an integer greater than or equal to 1. + +*name* is an identifier starting with a lowercase letter. *arity* is +an integer denoting the number of operands. + +*name* can optionally be preceded by `-` to indicate that it has been +obsoleted. The compiler is not allowed to generate BEAM files that +use obsolete instructions and the loader will refuse to load BEAM +files that use obsolete instructions. + +It only makes sense to define external generic instructions in the +file `genop.tab` in `lib/compiler/src`, because the compiler must +know about them in order to use them. + +New instructions must be added at the end of the file, with higher +numbers than the previous instructions. + +### Defining internal generic instructions ### + +Internal generic instructions are known only to the runtime +system and can be changed at any time without compatibility issues. + +There are two ways to define internal generic instructions: + +* Implicitly when a specific instruction is defined. This is by far +the most common way. Whenever a specific instruction is created, +**beam\_makeops** automatically creates an internal generic instruction +if it does not previously exist. + +* Explicitly. This is necessary only when a generic instruction does +not have any corresponding specific instruction. + +The syntax for an internal generic instruction is as follows: + +*name*/*arity* + +*name* is an identifier starting with a lowercase letter. *arity* is +an integer denoting the number of operands. + +### About generic instructions in general ### + +Each generic instruction has an opcode. The opcode is an integer, +greater than or equal to 1. For an external generic instruction, it +must be explicitly given `genop.tab`, while internal generic +instructions are automatically numbered by **beam\_makeops**. + +The identity of a generic instruction is its name combined with its +arity. That means that it is allowed to define two distinct generic +instructions having the same name but with different arities. For +example: + + move_window/5 + move_window/6 + +Each operand of a generic instruction is tagged with its type. A generic +instruction can have one of the following types: + +* `x` - X register. + +* `y` - Y register. + +* `l` - Floating point register number. + +* `i` - Tagged literal integer. + +* `a` - Tagged literal atom. + +* `n` - NIL (`[]`, the empty list). + +* `q` - Literal that don't fit in a word, that is an object stored on +the heap such as a list or tuple. Any heap object type is supported, +even types that don't have real literals such as external references. + +* `f` - Non-zero failure label. + +* `p` - Zero failure label. + +* `u` - Untagged integer that fits in a machine word. It is used for many +different purposes, such as the number of live registers in `test_heap/2`, +as a reference to the export for `call_ext/2`, and as the flags operand for +binary syntax instructions. When the generic instruction is translated to a +specific instruction, the type for the operand in the specific operation will +tell the loader how to treat the operand. + +* `o` - Overflow. If the value for an `u` operand does not fit in a machine +word, the type of the operand will be changed to `o` (with no associated +value). Currently only used internally in the loader in the guard constraint +function `binary_too_big()`. + +* `v` - Arity value. Only used internally in the loader. + + +### Defining specific instructions ### + +The specific instructions are known only to the runtime system and +are the instructions that are actually executed. They can be changed +at any time without causing compatibility issues. + +A specific instruction can have at most 6 operands. + +A specific instruction is defined by first giving its name followed by +the types for each operand. For example: + + move x y + +Internally, for example in the generated code and in the output from +the BEAM disassembler, the instruction `move x y` will be called `move_xy`. + +The name for a specific instruction is an identifier starting with a +lowercase letter. A type is an lowercase or uppercase letter. + +All specific instructions with a given name must have the same number +of operands. That is, the following is **not** allowed: + + move x x + move x y x y + +Here follows the type letters that more or less directly corresponds +to the types for generic instructions. + +* `x` - X register. Will be loaded as a byte offset to the X register +relative to the base of X register array. (Can be packed with other +operands.) + +* `y` - Y register. Will be loaded as a byte offset to the Y register +relative to the stack frame. (Can be packed with other operands.) + +* `r` - X register 0. An implicit operand that will not be stored in +the loaded code. + +* `l` - Floating point register number. (Can be packed with other +operands.) + +* `i` - Tagged literal integer (a SMALL that will fit in one word). + +* `a` - Tagged atom. + +* `n` - NIL or the empty list. (Will not be stored in the loaded code.) + +* `q` - Tagged CONS or BOXED pointer. That is, a term such as a list +or tuple. Any heap object type is supported, even types that don't +have real literals such as external references. + +* `f` - Failure label (non-zero). The target for a branch +or call instruction. + +* `p` - The 0 failure label, meaning that an exception should be raised +if the instruction fails. (Will not be stored in the loaded code.) + +* `c` - Any literal term; that is, immediate literals such as SMALL, +and CONS or BOXED pointers to literals. (Can be used where the +operand in the generic instruction has one of the types `i`, `a`, `n`, +or `q`.) + +The types that follow do a type test of the operand at runtime; thus, +they are generally more expensive in terms of runtime than the types +described earlier. However, those operand types are needed to avoid a +combinatorial explosion in the number of specific instructions and +overall code size of `process_main()`. + +* `s` - Tagged source: X register, Y register, or a literal term. The +tag will be tested at runtime to retrieve the value from an X +register, a Y register, or simply use the value as a tagged Erlang +term. (Implementation note: An X register is tagged as a pid, and a Y +register as a port. Therefore the literal term must not contain a +port or pid.) + +* `S` - Tagged source register (X or Y). The tag will be tested at +runtime to retrieve the value from an X register or a Y register. Slighly +cheaper than `s`. + +* `d` - Tagged destination register (X or Y). The tag will be tested +at runtime to set up a pointer to the destination register. If the +instrution performs a garbarge collection, it must use the +`$REFRESH_GEN_DEST()` macro to refresh the pointer before storing to +it (there are more details about that in a later section). + +* `j` - A failure label (combination of `f` and `p`). If the branch target 0, +an exception will be raised if instruction fails, otherwise control will be +transfered to the target address. + +The types that follows are all applied to an operand that has the `u` +type. + +* `t` - An untagged integer that will fit in 12 bits (0-4096). It can be +packed with other operands in a word. Most often used as the number +of live registers in instructions such as `test_heap`. + +* `I` - An untagged integer that will fit in 32 bits. It can be +packed with other operands in a word on a 64-bit system. + +* `W` - Untagged integer or pointer. Not possible to pack with other +operands. + +* `e` - Pointer to an export entry. Use by call instructions that call +other modules, such as `call_ext`. + +* `L` - A label. Only used by the `label/1` instruction. + +* `b` - Pointer to BIF. Used by instructions that BIFs, such as +`call_bif`. + +* `A` - A tagged arityvalue. Used in instructions that test the arity +of a tuple. + +* `P` - A byte offset into a tuple. + +* `Q` - A byte offset into the stack. Used for updating the frame +pointer register. Can be packed with other operands. + +When the loader translates a generic instruction a specific +instruction, it will choose the most specific instruction that will +fit the types. Consider the following two instructions: + + move c x + move n x + +The `c` operand can encode any literal value, including NIL. The +`n` operand only works for NIL. If we have the generic instruction +`{move,nil,{x,1}}`, the loader will translate it to `move_nx 1` +because `move n x` is more specific. `move_nx` could be slightly +faster or smaller (depending on the architecture), because the `[]` +is not stored explicitly as an operand. + +#### Syntactic sugar for specific instructions #### + +It is possible to specify more than one type letter for each operand. +Here is an example: + + move cxy xy + +This is syntactic sugar for: + + move c x + move c y + move x x + move x y + move y x + move y y + +Note the difference between `move c xy` and `move c d`. Note that `move c xy` +is equivalent to the following two definitions: + + move c x + move c y + +On the other hand, `move c d` is a single instruction. At runtime, +the `d` operand will be tested to see whether it refers to an X +register or a Y register, and a pointer to the register will be set +up. + +#### The '?' type modifier #### + +The character `?` can be added to the end of an operand to indicate +that the operand will not be used every time the instruction is executed. +For example: + + allocate_heap t I t? + is_eq_exact f? x xy + +In `allocate_heap`, the last operand is the number of live registers. +It will only be used if there is not enough heap space and a garbage +collection must be performed. + +In `is_eq_exact`, the failure address (the first operand) will only be +used if the two register operands are not equal. + +Knowing that an operand is not always used can improve how packing +is done for some instructions. + +For the `allocate_heap` instruction, without the `?` the packing would +be done like this: + + +--------------------+--------------------+ + I -> | Stack needed | &&lb_allocate_heap + + +--------------------+--------------------+ + | Heap needed | Live registers + + +--------------------+--------------------+ + +"Stack needed" and "Heap needed" are always used, but they are in +different words. Thus, at runtime the `allocate_heap` instruction +must read both words from memory even though it will not always use +"Live registers". + +With the `?`, the operands will be packed like this: + + +--------------------+--------------------+ + I -> | Live registers | &&lb_allocate_heap + + +--------------------+--------------------+ + | Heap needed | Stack needed + + +--------------------+--------------------+ + +Now "Stack needed" and "Heap needed" are in the same word. + +### Defining transformation rules ### + +Transformation rules are used to rewrite generic instructions to other +generic instructions. The transformations rules are applied +repeatedly until no rule match. At that point, the first instruction +in the resulting instruction sequence will be converted to a specific +instruction and added to the code for the module being loaded. Then +the transformation rules for the remaining instructions are run in the +same way. + +A rule is recognized by its right-pointer arrow: `=>`. To the left of +the arrow is one or more instruction patterns, separated by `|`. To +the right of the arrow is zero or more instructions, separated by `|`. +If the instructions from the BEAM code matches the instruction +patterns on the left side, they will be replaced with instructions on +the right side (or removed if there are no instructions on the right). + +#### Defining instruction patterns #### + +We will start looking at the patterns on the left side of the arrow. + +A pattern for an instruction consists of its name, followed by a pattern +for each of its operands. The operand patterns are separated by spaces. + +The simplest possible pattern is a variable. Just like in Erlang, +a variable must begin with an uppercase letter. If the same variable is +used in multiple operands, the pattern will only match if the operands +are equal. For example: + + move Same Same => + +This pattern will match if the operands for `move` are the same. If +the pattern match, the instruction will be removed. (That used to be an +actual rule a long time ago when the compiler would occasionally produce +instructions such as `{move,{x,2},{x,2}}`.) + +Variables that have been bound on the left side can be used on the +right side. For example, this rule will rewrite all `move` instructions +to `assign` instructions with the operands swapped: + + move Src Dst => assign Dst Src + +If we only want to match operands of a certain type, we can +use a type constraint. A type constraint consists of one or more +lowercase letters, each specifying a type. For example: + + is_integer Fail an => jump Fail + +The second operand pattern, `an`, will match if the second operand is +either an atom or NIL (the empty list). In case of a match, the +`is_integer/2` instruction will be replaced with a `jump/1` +instruction. + +An operand pattern can bind a variable and constrain the type at the +same time by following the variable with a `=` and the constraint. +For example: + + is_eq_exact Fail=f R=xy C=q => i_is_eq_exact_literal Fail R C + +Here the `is_eq_exact` instruction is replaced with a specialized instruction +that only compares literals, but only if the first operand is a register and +the second operand is a literal. + +#### Further constraining patterns #### + +In addition to specifying a type letter, the actual value for the type can +be specified. For example: + + move C=c x==1 => move_x1 C + +Here the second operand of `move` is constrained to be X register 1. + +When specifying an atom constraint, the atom is written as it would be +in the C source code. That is, it needs an `am_` prefix, and it must +be listed in `atom.names`. For example: + + is_boolean Fail=f a==am_true => + is_boolean Fail=f a==am_false => + +There are several constraints available for testing whether a call is to a BIF +or a function. + +The constraint `u$is_bif` will test whether the given operand refers to a BIF. +For example: + + call_ext u Bif=u$is_bif => call_bif Bif + call_ext u Func => i_call_ext Func + +The `call_ext` instruction can be used to call functions written in +Erlang as well as BIFs (or more properly called SNIFs). The +`u$is_bif` constraint will match if the operand refers to a BIF (that +is, if it is listed in the file `bif.tab`). Note that `u$is_bif` +should only be applied to operands that are known to contain an index +to the import table chunk in the BEAM file (such operands have the +type `b` or `e` in the corresponding specific instruction). If +applied to other `u` operands, it will at best return a nonsense +result. + +The `u$is_not_bif` constraint matches if the operand does not refer to +a BIF (not listed in `bif.tab`). For example: + + move S X0=x==0 | line Loc | call_ext_last Ar Func=u$is_not_bif D => \ + move S X0 | call_ext_last Ar Func D + +The `u$bif:Module:Name/Arity` constraint tests whether the given +operand refers to a specific BIF. Note that `Module:Name/Arity` +**must** be an existing BIF defined in `bif.tab`, or there will +be a compilation error. It is useful when a call to a specific BIF +should be replaced with an instruction as in this example: + + gc_bif2 Fail Live u$bif:erlang:splus/2 S1 S2 Dst => \ + gen_plus Fail Live S1 S2 Dst + +Here the call to the GC BIF `'+'/2` will be replaced with the instruction +`gen_plus/5`. Note that the same name as used in the C source code must be +used for the BIF, which in this case is `splus`. It is defined like this +in `bit.tab`: + + ubif erlang:'+'/2 splus_2 + +The `u$func:Module:Name/Arity` will test whether the given operand is a +a specific function. Here is an example: + + bif1 Fail u$func:erlang:is_constant/1 Src Dst => too_old_compiler + +`is_constant/1` used to be a BIF a long time ago. The transformation +replaces the call with the `too_old_compiler` instruction which will produce +a nicer error message than the default error would be for a missing guard BIF. + +#### Type constraints allowed in patterns #### + +Here are all type letters that are allowed on the left side of a transformation +rule. + +* `u` - An untagged integer that fits in a machine word. + +* `x` - X register. + +* `y` - Y register. + +* `l` - Floating point register number. + +* `i` - Tagged literal integer. + +* `a` - Tagged literal atom. + +* `n` - NIL (`[]`, the empty list). + +* `q` - Literals that don't fit in a word, such as list or tuples. + +* `f` - Non-zero failure label. + +* `p` - The zero failure label. + +* `j` - Any label. Equivalent to `fp`. + +* `c` - Any literal term. Equivalent to `ainq`. + +* `s` - X register, Y register, or any literal term. Equivalent to `xyc`. + +* `d` - X or Y register. Equivalent to `xy`. (In a pattern `d` will +match both source and destination registers. As an operand in a specific +instruction, it must only be used for a destination register.) + +* `o` - Overflow. An untagged integer that does not fit in a machine word. + +#### Guard constraints #### + +If the constraints described so far is not enough, additional +constraints can be written in C in `beam_load.c` and be called as a +guard function on the left side of the transformation. If the guard +function returns a non-zero value, the matching of the rule will +continue, otherwise the match will fail. For example: + + ensure_map Lit=q | literal_is_map(Lit) => + +The guard test `literal_is_map/1` tests whether the given literal is a map. +If the literal is a map, the instruction is unnecessary and can be removed. + +It is outside the scope for this document to describe in detail how such +guard functions are written, but for the curious here is the implementation +of `literal_is_map()`: + + static int + literal_is_map(LoaderState* stp, GenOpArg Lit) + { + Eterm term; + + ASSERT(Lit.type == TAG_q); + term = stp->literals[Lit.val].term; + return is_map(term); + } + +#### Handling instruction with variable number of operands #### + +Some instructions, such as `select_val/3`, essentially has a variable +number of operands. Such instructions have a `{list,[...]}` operand +as their last operand in the BEAM assembly code. For example: + + {select_val,{x,0}, + {f,1}, + {list,[{atom,b},{f,4},{atom,a},{f,5}]}}. + +The loader will convert a `{list,[...]}` operand to an `u` operand whose +value is the number of elements in the list, followed by each element in +the list. The instruction above would be translated to the following +generic instruction: + + {select_val,{x,0},{f,1},{u,4},{atom,b},{f,4},{atom,a},{f,5}} + +To match a variable number of arguments we need to use the special +operand type `*` like this: + + select_val Src=aiq Fail=f Size=u List=* => \ + i_const_select_val Src Fail Size List + +This transformation renames a `select_val/3` instruction +with a constant source operand to `i_const_select_val/3`. + +#### Constructing new instructions on the right side #### + +The most common operand on the right side is a variable that was bound while +matching the left side. For example: + + trim N Remaining => i_trim N + +An operand can also be a type letter to construct an operand of that type. +Each type has a default value. For example, the type `x` has the default +value 1023, which is the highest X register. That makes `x` on the right +side a convenient shortcut for a temporary X register. For example: + + is_number Fail Literal=q => move Literal x | is_number Fail x + +If the second operand for `is_number/2` is a literal, it will be moved to +X register 1023. Then `is_number/2` will test whether the value stored in +X register 1023 is a number. + +This kind of transformation is useful when it is rare that an operand can +be anything else but a register. In the case of `is_number/2`, the second +operand is always a register unless the compiler optimizations have been +disabled. + +If the default value is not suitable, the type letter can be followed +by `=` and a value. Most types take an integer value. The value for +an atom is written the same way as in the C source code. For example, +the atom `false` is written as `am_false`. The atom must be listed in +`atom.names`. + +Here is an example showing how values can be specified: + + bs_put_utf32 Fail=j Flags=u Src=s => \ + i_bs_validate_unicode Fail Src | \ + bs_put_integer Fail i=32 u=1 Flags Src + +#### Type letters on the right side #### + +Here follows all types that are allowed to be used in operands for +instructions being constructed on the right side of a transformation +rule. + +* `u` - Construct an untagged integer. The default value is 0. + +* `x` - X register. The default value is 1023. That makes `x` convenient to +use as a temporary X register. + +* `y` - Y register. The default value is 0. + +* `l` - Foating point register number. The default value is 0. + +* `i` - Tagged literal integer. The default value is 0. + +* `a` - Tagged atom. The default value is the empty atom (`am_Empty`). + +* `n` - NIL (`[]`, the empty list). + +#### Function call on the right side #### + +Transformations that are not possible to describe with the rule +language as described here can be written as a C function in +`beam_load.c` and called from the right side of a transformation. The +left side of the transformation will perform the match and bind +operands to variables. The variables can then be passed to a +generator function on the right side. For example: + + bif2 Fail=j u$bif:erlang:element/2 Index=s Tuple=xy Dst=d => \ + gen_element(Jump, Index, Tuple, Dst) + +This transformation rule matches a call to the BIF `element/2`. +The operands will be captured and the function `gen_element()` will +be called. + +`gen_element()` will produce one of two instructions depending +on `Index`. If `Index` is an integer in the range from 1 up to +the maximum tuple size, the instruction `i_fast_element/2` will +be produced, otherwise the instruction `i_element/4` will be +produced. The corresponding specific instructions are: + + i_fast_element xy j? I d + i_element xy j? s d + +The `i_fast_element/2` instruction is faster because the tuple is +already an untagged integer. It also knows that the index is at least +1, so it does not have to test for that. The `i_element/4` +instruction will have to fetch the index from a register, test that it +is an integer, and untag the integer. + +It is outside the scope of this document to describe in detail how +generator functions are written, but for the curious, here is the +implementation of `gen_element()`: + + static GenOp* + gen_element(LoaderState* stp, GenOpArg Fail, + GenOpArg Index, GenOpArg Tuple, GenOpArg Dst) + { + GenOp* op; + + NEW_GENOP(stp, op); + op->arity = 4; + op->next = NULL; + + if (Index.type == TAG_i && Index.val > 0 && + Index.val <= ERTS_MAX_TUPLE_SIZE && + (Tuple.type == TAG_x || Tuple.type == TAG_y)) { + op->op = genop_i_fast_element_4; + op->a[0] = Tuple; + op->a[1] = Fail; + op->a[2].type = TAG_u; + op->a[2].val = Index.val; + op->a[3] = Dst; + } else { + op->op = genop_i_element_4; + op->a[0] = Tuple; + op->a[1] = Fail; + op->a[2] = Index; + op->a[3] = Dst; + } + + return op; + } +} + +### Defining the implementation ### + +The actual implementation of instructions are also defined in `.tab` +files processed by **beam\_makeops**. For practical reasons, +instruction definitions are stored in several files, at the time of +writing in the following files: + + bif_instrs.tab + arith_instrs.tab + bs_instrs.tab + float_instrs.tab + instrs.tab + map_instrs.tab + msg_instrs.tab + select_instrs.tab + trace_instrs.tab + +There is also a file that only contains macro definitions: + + macros.tab + +The syntax of each file is similar to C code. In fact, most of +the contents *is* C code, interspersed with macro invocations. + +To allow Emacs to auto-indent the code, each file starts with the +following line: + + // -*- c -*- + +To avoid messing up the indentation, all comments are written +as C++ style comments (`//`) instead of `#`. Note that a comment +must start at the beginning of a line. + +The meat of an instruction definition file are macro definitions. +We have seen this macro definition before: + + move(Src, Dst) { + $Dst = $Src; + } + +A macro definitions must start at the beginning of the line (no spaces +allowed), the opening curly bracket must be on the same line, and the +finishing curly bracket must be at the beginning of a line. It is +recommended that the macro body is properly indented. + +As a convention, the macro arguments in the head all start with an +uppercase letter. In the body, the macro arguments can be expanded +by preceding them with `$`. + +A macro definition whose name and arity matches a family of +specific instructions is assumed to be the implementation of that +instruction. + +A macro can also be invoked from within another macro. For example, +`move_deallocate_return/2` avoids repeating code by invoking +`$deallocate_return()` as a macro: + + move_deallocate_return(Src, Deallocate) { + x(0) = $Src; + $deallocate_return($Deallocate); + } + +Here is the definition of `deallocate_return/1`: + + deallocate_return(Deallocate) { + //| -no_next + int words_to_pop = $Deallocate; + SET_I((BeamInstr *) cp_val(*E)); + E = ADD_BYTE_OFFSET(E, words_to_pop); + CHECK_TERM(x(0)); + DispatchReturn; + } + +The expanded code for `move_deallocate_return` will look this: + + OpCase(move_deallocate_return_cQ): + { + x(0) = I[1]; + do { + int words_to_pop = Qb(BeamExtraData(I[0])); + SET_I((BeamInstr *) cp_val(*E)); + E = ADD_BYTE_OFFSET(E, words_to_pop); + CHECK_TERM(x(0)); + DispatchReturn; + } while (0); + } + +When expanding macros, **beam\_makeops** wraps the expansion in a +`do`/`while` wrapper unless **beam\_makeops** can clearly see that no +wrapper is needed. In this case, the wrapper is needed. + +Note that arguments for macros cannot be complex expressions, because +the arguments are split on `,`. For example, the following would +not work because **beam\_makeops** would split the expression into +two arguments: + + $deallocate_return(get_deallocation(y, $Deallocate)); + +#### Code generation directives #### + +Within macro definitions, `//` comments are in general not treated +specially. They will be copied to the file with the generated code +along with the rest of code in the body. + +However, there is an exception. Within a macro definition, a line that +starts with whitespace followed by `//|` is treated specially. The +rest of the line is assumed to contain directives to control code +generation. + +Currently, two code generation directives are recognized: + +* `-no_prefetch` +* `-no_next` + +##### The -no_prefetch directive ##### + +To see what `-no_prefetch` does, let's first look at the default code +generation. Here is the code generated for `move_cx`: + + OpCase(move_cx): + { + BeamInstr next_pf = BeamCodeAddr(I[2]); + xb(BeamExtraData(I[0])) = I[1]; + I += 2; + ASSERT(VALID_INSTR(next_pf)); + GotoPF(next_pf); + } + +Note that the very first thing done is to fetch the address to the +next instruction. The reason is that it usually improves performance. + +Just as a demonstration, we can add a `-no_prefetch` directive to +the `move/2` instruction: + + move(Src, Dst) { + //| -no_prefetch + $Dst = $Src; + } + +We can see that the prefetch is no longer done: + + OpCase(move_cx): + { + xb(BeamExtraData(I[0])) = I[1]; + I += 2; + ASSERT(VALID_INSTR(*I)); + Goto(*I); + } + +When would we want to turn off the prefetch in practice? + +In instructions that will not always execute the next instruction. +For example: + + is_atom(Fail, Src) { + if (is_not_atom($Src)) { + $FAIL($Fail); + } + } + + // From macros.tab + FAIL(Fail) { + //| -no_prefetch + $SET_I_REL($Fail); + Goto(*I); + } + +`is_atom/2` may either execute the next instruction (if the second +operand is an atom) or branch to the failure label. + +The generated code looks like this: + + OpCase(is_atom_fx): + { + if (is_not_atom(xb(I[1]))) { + ASSERT(VALID_INSTR(*(I + (fb(BeamExtraData(I[0]))) + 0))); + I += fb(BeamExtraData(I[0])) + 0;; + Goto(*I);; + } + I += 2; + ASSERT(VALID_INSTR(*I)); + Goto(*I); + } + +##### The -no_next directive ##### + +Next we will look at when the `-no_next` directive can be used. Here +is the `jump/1` instruction: + + jump(Fail) { + $JUMP($Fail); + } + + // From macros.tab + JUMP(Fail) { + //| -no_next + $SET_I_REL($Fail); + Goto(*I); + } + +The generated code looks like this: + + OpCase(jump_f): + { + ASSERT(VALID_INSTR(*(I + (fb(BeamExtraData(I[0]))) + 0))); + I += fb(BeamExtraData(I[0])) + 0;; + Goto(*I);; + } + +If we remove the `-no_next` directive, the code would look like this: + + OpCase(jump_f): + { + BeamInstr next_pf = BeamCodeAddr(I[1]); + ASSERT(VALID_INSTR(*(I + (fb(BeamExtraData(I[0]))) + 0))); + I += fb(BeamExtraData(I[0])) + 0;; + Goto(*I);; + I += 1; + ASSERT(VALID_INSTR(next_pf)); + GotoPF(next_pf); + } + +In the end, the C compiler will probably optimize this code to the +same native code as the first version, but the first version is certainly +much easier to read for human readers. + +#### Macros in the macros.tab file #### + +The file `macros.tab` contains many useful macros. When implementing +new instructions it is good practice to look through `macros.tab` to +see if any of existing macros can be used rather than re-inventing +the wheel. + +We will describe a few of the most useful macros here. + +##### The GC_REGEXP definition ##### + +The following line defines a regular expression that will recognize +a call to a function that does a garbage collection: + + GC_REGEXP=erts_garbage_collect|erts_gc|GcBifFunction; + +The purpose is that **beam\_makeops** can verify that an instruction +that does a garbage collection and has an `d` operand uses the +`$REFRESH_GEN_DEST()` macro. + +If you need to define a new function that does garbage collection, +you should give it the prefix `erts_gc_`. If that is not possible +you should update the regular expression so that it will match your +new function. + +##### FAIL(Fail) ##### + +Branch to `$Fail`. Will suppress prefetch (`-no_prefetch`). Typical use: + + is_nonempty_list(Fail, Src) { + if (is_not_list($Src)) { + $FAIL($Fail); + } + } + +##### JUMP(Fail) ##### + +Branch to `$Fail`. Suppresses generation of dispatch of the next +instruction (`-no_next`). Typical use: + + jump(Fail) { + $JUMP($Fail); + } + +##### GC_TEST(NeedStack, NeedHeap, Live) ##### + +`$GC_TEST(NeedStack, NeedHeap, Live)` tests that given amount of +stack space and heap space is available. If not it will do a +garbage collection. Typical use: + + test_heap(Nh, Live) { + $GC_TEST(0, $Nh, $Live); + } + +##### AH(NeedStack, NeedHeap, Live) ##### + +`AH(NeedStack, NeedHeap, Live)` allocates a stack frame and +optionally additional heap space. + +#### Pre-defined macros and variables #### + +**beam\_makeops** defines several built-in macros and pre-bound variables. + +##### The NEXT_INSTRUCTION pre-bound variable ##### + +The NEXT_INSTRUCTION is a pre-bound variable that is available in +all instructions. It expands to the address of the next instruction. + +Here is an example: + + i_call(CallDest) { + SET_CP(c_p, $NEXT_INSTRUCTION); + $DISPATCH_REL($CallDest); + } + +When calling a function, the return address is first stored in `c_p->cp` +(using the `SET_CP()` macro defined in `beam_emu.c`), and then control is +transferred to the callee. Here is the generated code: + + OpCase(i_call_f): + { + SET_CP(c_p, I+1); + ASSERT(VALID_INSTR(*(I + (fb(BeamExtraData(I[0]))) + 0))); + I += fb(BeamExtraData(I[0])) + 0;; + DTRACE_LOCAL_CALL(c_p, erts_code_to_codemfa(I)); + Dispatch();; + } + +We can see that that `$NEXT_INSTRUCTION` has been expanded to `I+1`. +That makes sense since the size of the `i_call_f/1` instruction is +one word. + +##### The IP_ADJUSTMENT pre-bound variable ##### + +`$IP_ADJUSTMENT` is usually 0. In a few combined instructions +(described below) it can be non-zero. It is used like this +in `macros.tab`: + + SET_I_REL(Offset) { + ASSERT(VALID_INSTR(*(I + ($Offset) + $IP_ADJUSTMENT))); + I += $Offset + $IP_ADJUSTMENT; + } + +Avoid using `IP_ADJUSTMENT` directly. Use `SET_I_REL()` or +one of the macros that invoke such as `FAIL()` or `JUMP()` +defined in `macros.tab`. + +#### Pre-defined macro functions #### + +##### The IF() macro ##### + +`$IF(Expr, IfTrue, IfFalse)` evaluates `Expr`, which must be a valid +Perl expression (which for simple numeric expressions have the same +syntax as C). If `Expr` evaluates to 0, the entire `IF()` expression will be +replaced with `IfFalse`, otherwise it will be replaced with `IfTrue`. + +See the description of `OPERAND_POSITION()` for an example. + +##### The OPERAND\_POSITION() macro ##### + +`$OPERAND_POSITION(Expr)` returns the position for `Expr`, if +`Expr` is an operand that is not packed. The first operand is +at position 1. + +Returns 0 otherwise. + +This macro could be used like this in order to share code: + + FAIL(Fail) { + //| -no_prefetch + $IF($OPERAND_POSITION($Fail) == 1 && $IP_ADJUSTMENT == 0, + goto common_jump, + $DO_JUMP($Fail)); + } + + DO_JUMP(Fail) { + $SET_I_REL($Fail); + Goto(*I)); + } + + // In beam_emu.c: + common_jump: + I += I[1]; + Goto(*I)); + + +#### The $REFRESH\_GEN\_DEST() macro #### + +When a specific instruction has a `d` operand, early during execution +of the instruction, a pointer will be initialized to point to the X or +Y register in question. + +If there is a garbage collection before the result is stored, +the stack will move and if the `d` operand refered to a Y +register, the pointer will no longer be valid. (Y registers are +stored on the stack.) + +In those circumstances, `$REFRESH_GEN_DEST()` must be invoked +to set up the pointer again. **beam\_makeops** will notice +if there is a call to a function that does a garbage collection and +`$REFRESH_GEN_DEST()` is not called. + +Here is a complete example. The `new_map` instruction is defined +like this: + + new_map d t I + +It is implemented like this: + + new_map(Dst, Live, N) { + Eterm res; + + HEAVY_SWAPOUT; + res = erts_gc_new_map(c_p, reg, $Live, $N, $NEXT_INSTRUCTION); + HEAVY_SWAPIN; + $REFRESH_GEN_DEST(); + $Dst = res; + $NEXT($NEXT_INSTRUCTION+$N); + } + +If we have forgotten the `$REFRESH_GEN_DEST()` there would be a message +similar to this: + + pointer to destination register is invalid after GC -- use $REFRESH_GEN_DEST() + ... from the body of new_map at beam/map_instrs.tab(30) + +#### Combined instructions #### + +**Problem**: For frequently executed instructions we want to use +"fast" operands types such as `x` and `y`, as opposed to `s` or `S`. +To avoid an explosion in code size, we want to share most of the +implementation between the instructions. Here are the specific +instructions for `i_increment/5`: + + i_increment r W t d + i_increment x W t d + i_increment y W t d + +The `i_increment` instruction is implemented like this: + + i_increment(Source, IncrementVal, Live, Dst) { + Eterm increment_reg_source = $Source; + Eterm increment_val = $IncrementVal; + Uint live; + Eterm result; + + if (ERTS_LIKELY(is_small(increment_reg_val))) { + Sint i = signed_val(increment_reg_val) + increment_val; + if (ERTS_LIKELY(IS_SSMALL(i))) { + $Dst = make_small(i); + $NEXT0(); + } + } + live = $Live; + HEAVY_SWAPOUT; + reg[live] = increment_reg_val; + reg[live+1] = make_small(increment_val); + result = erts_gc_mixed_plus(c_p, reg, live); + HEAVY_SWAPIN; + ERTS_HOLE_CHECK(c_p); + if (ERTS_LIKELY(is_value(result))) { + $REFRESH_GEN_DEST(); + $Dst = result; + $NEXT0(); + } + ASSERT(c_p->freason != BADMATCH || is_value(c_p->fvalue)); + goto find_func_info; + } + +There will be three almost identical copies of the code. Given the +size of the code, that could be too high cost to pay. + +To avoid the three copies of the code, we could use only one specific +instruction: + + i_increment S W t d + +(The same implementation as above will work.) + +That reduces the code size, but is slower because `S` means that +there will be extra code to test whether the operand refers to an X +register or a Y register. + +**Solution**: We can use "combined instructions". Combined +instructions are combined from instruction fragments. The +bulk of the code can be shared. + +Here we will show how `i_increment` can be implemented as a combined +instruction. We will show each individual fragment first, and then +show how to connect them together. First we will need a variable that +we can store the value fetched from the register in: + + increment.head() { + Eterm increment_reg_val; + } + +The name `increment` is the name of the group that the fragment +belongs to. Note that it does not need to have the same +name as the instruction. The group name is followed by `.` and +the name of the fragment. The name `head` is pre-defined. +The code in it will be placed at the beginning of a block, so +that all fragments in the group can access it. + +Next we define the fragment that will pick up the value from the +register from the first operand: + + increment.fetch(Src) { + increment_reg_val = $Src; + } + +We call this fragment `fetch`. This fragment will be duplicated three +times, one for each value of the first operand (`r`, `x`, and `y`). + +Next we define the main part of the code that do the actual incrementing. + + increment.execute(IncrementVal, Live, Dst) { + Eterm increment_val = $IncrementVal; + Uint live; + Eterm result; + + if (ERTS_LIKELY(is_small(increment_reg_val))) { + Sint i = signed_val(increment_reg_val) + increment_val; + if (ERTS_LIKELY(IS_SSMALL(i))) { + $Dst = make_small(i); + $NEXT0(); + } + } + live = $Live; + HEAVY_SWAPOUT; + reg[live] = increment_reg_val; + reg[live+1] = make_small(increment_val); + result = erts_gc_mixed_plus(c_p, reg, live); + HEAVY_SWAPIN; + ERTS_HOLE_CHECK(c_p); + if (ERTS_LIKELY(is_value(result))) { + $REFRESH_GEN_DEST(); + $Dst = result; + $NEXT0(); + } + ASSERT(c_p->freason != BADMATCH || is_value(c_p->fvalue)); + goto find_func_info; + } + +We call this fragment `execute`. It will handle the three remaining +operands (`W t d`). There will only be one copy of this fragment. + +Now that we have defined the fragments, we need to inform +**beam\_makeops** how they should be connected: + + i_increment := increment.fetch.execute; + +To the left of the `:=` is the name of the specific instruction that +should be implemented by the fragments, in this case `i_increment`. +To the right of `:=` is the name of the group with the fragments, +followed by a `.`. Then the name of the fragments in the group are +listed in the order they should be executed. Note that the `head` +fragment is not listed. + +The line ends in `;` (to avoid messing up the indentation in Emacs). + +(Note that in practice the `:=` line is usually placed before the +fragments.) + +The generated code looks like this: + + { + Eterm increment_reg_val; + OpCase(i_increment_rWtd): + { + increment_reg_val = r(0); + } + goto increment__execute; + + OpCase(i_increment_xWtd): + { + increment_reg_val = xb(BeamExtraData(I[0])); + } + goto increment__execute; + + OpCase(i_increment_yWtd): + { + increment_reg_val = yb(BeamExtraData(I[0])); + } + goto increment__execute; + + increment__execute: + { + // Here follows the code from increment.execute() + . + . + . + } + +##### Some notes about combined instructions ##### + +The operands that are different must be at +the beginning of the instruction. All operands in the last +fragment must have the same operands in all variants of +the specific instruction. + +As an example, the following specific instructions cannot be +implemented as a combined instruction: + + i_times j? t x x d + i_times j? t x y d + i_times j? t s s d + +We would have to change the order of the operands so that the +two operands that are different are placed first: + + i_times x x j? t d + i_times x y j? t d + i_times s s j? t d + +We can then define: + + i_times := times.fetch.execute; + + times.head { + Eterm op1, op2; + } + + times.fetch(Src1, Src2) { + op1 = $Src1; + op2 = $Src2; + } + + times.execute(Fail, Live, Dst) { + // Multiply op1 and op2. + . + . + . + } + +Several instructions can share a group. As an example, the following +instructions have different names, but in the end they all create a +binary. The last two operands are common for all of them: + + i_bs_init_fail xy j? t? x + i_bs_init_fail_heap s I j? t? x + i_bs_init W t? x + i_bs_init_heap W I t? x + +The instructions are defined like this (formatted with extra +spaces for clarity): + + i_bs_init_fail_heap := bs_init . fail_heap . verify . execute; + i_bs_init_fail := bs_init . fail . verify . execute; + i_bs_init := bs_init . . plain . execute; + i_bs_init_heap := bs_init . heap . execute; + +Note that the first two instruction have three fragments, while the +other two only have two fragments. Here are the fragments: + + bs_init_bits.head() { + Eterm num_bits_term; + Uint num_bits; + Uint alloc; + } + + bs_init_bits.plain(NumBits) { + num_bits = $NumBits; + alloc = 0; + } + + bs_init_bits.heap(NumBits, Alloc) { + num_bits = $NumBits; + alloc = $Alloc; + } + + bs_init_bits.fail(NumBitsTerm) { + num_bits_term = $NumBitsTerm; + alloc = 0; + } + + bs_init_bits.fail_heap(NumBitsTerm, Alloc) { + num_bits_term = $NumBitsTerm; + alloc = $Alloc; + } + + bs_init_bits.verify(Fail) { + // Verify the num_bits_term, fail using $FAIL + // if there is a problem. + . + . + . + } + + bs_init_bits.execute(Live, Dst) { + // Long complicated code to a create a binary. + . + . + . + } + +The full definitions of those instructions can be found in `bs_instrs.tab`. +The generated code can be found in `beam_warm.h`. diff --git a/erts/emulator/nifs/common/zlib_nif.c b/erts/emulator/nifs/common/zlib_nif.c index fa29b4fb71..b709ed5a6f 100644 --- a/erts/emulator/nifs/common/zlib_nif.c +++ b/erts/emulator/nifs/common/zlib_nif.c @@ -717,7 +717,9 @@ static ERL_NIF_TERM zlib_deflateEnd(ErlNifEnv *env, int argc, const ERL_NIF_TERM static ERL_NIF_TERM zlib_deflateParams(ErlNifEnv *env, int argc, const ERL_NIF_TERM argv[]) { zlib_data_t *d; + int res, level, strategy; + Bytef dummy_buffer; if(argc != 3 || !get_zlib_data(env, argv[0], &d) || !enif_get_int(env, argv[1], &level) @@ -729,12 +731,27 @@ static ERL_NIF_TERM zlib_deflateParams(ErlNifEnv *env, int argc, const ERL_NIF_T return enif_raise_exception(env, am_not_initialized); } - /* deflateParams will flush everything currently in the stream, corrupting - * the heap unless it's empty. We therefore pretend to have a full output - * buffer, forcing a Z_BUF_ERROR if there's anything left to be flushed. */ - d->s.avail_out = 0; + /* This is a bit of a hack; deflateParams flushes with Z_BLOCK which won't + * stop at a byte boundary, so we can't split this operation up, and we + * can't allocate a buffer large enough to fit it in one go since we have + * to support zlib versions that lack deflatePending. + * + * We therefore flush everything prior to this call to ensure that we are + * stopped on a byte boundary and have no pending data. We then hand it a + * dummy buffer to detect when this assumption doesn't hold (Hopefully + * never), and to smooth over an issue with zlib 1.2.11 which always + * returns Z_BUF_ERROR when d->s.avail_out is 0, regardless of whether + * there's any pending data or not. */ + + d->s.next_out = &dummy_buffer; + d->s.avail_out = 1; + res = deflateParams(&d->s, level, strategy); + if(d->s.avail_out == 0) { + return zlib_return(env, Z_STREAM_ERROR); + } + return zlib_return(env, res); } @@ -929,7 +946,7 @@ static ERL_NIF_TERM zlib_inflate(ErlNifEnv *env, int argc, const ERL_NIF_TERM ar return enif_raise_exception(env, am_not_initialized); } - if(d->eos_seen) { + if(d->eos_seen && enif_ioq_size(d->input_queue) > 0) { int res; switch(d->eos_behavior) { @@ -943,11 +960,10 @@ static ERL_NIF_TERM zlib_inflate(ErlNifEnv *env, int argc, const ERL_NIF_TERM ar } d->eos_seen = 0; + break; case EOS_BEHAVIOR_CUT: zlib_reset_input(d); - - return enif_make_tuple2(env, am_finished, enif_make_list(env, 0)); } } diff --git a/erts/emulator/test/distribution_SUITE.erl b/erts/emulator/test/distribution_SUITE.erl index 2d0ae9c83e..e2914cbc92 100644 --- a/erts/emulator/test/distribution_SUITE.erl +++ b/erts/emulator/test/distribution_SUITE.erl @@ -1365,81 +1365,59 @@ bad_dist_structure(Config) when is_list(Config) -> start_monitor(Offender,P), P ! one, send_bad_structure(Offender, P,{?DOP_MONITOR_P_EXIT,'replace',P,normal},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_monitor(Offender,P), send_bad_structure(Offender, P,{?DOP_MONITOR_P_EXIT,'replace',P,normal,normal},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_link(Offender,P), send_bad_structure(Offender, P,{?DOP_LINK},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_link(Offender,P), send_bad_structure(Offender, P,{?DOP_UNLINK,'replace'},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_link(Offender,P), send_bad_structure(Offender, P,{?DOP_UNLINK,'replace',make_ref()},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_link(Offender,P), send_bad_structure(Offender, P,{?DOP_UNLINK,make_ref(),P},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_link(Offender,P), send_bad_structure(Offender, P,{?DOP_UNLINK,normal,normal},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_monitor(Offender,P), send_bad_structure(Offender, P,{?DOP_MONITOR_P,'replace',P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_monitor(Offender,P), send_bad_structure(Offender, P,{?DOP_MONITOR_P,'replace',P,normal},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_monitor(Offender,P), send_bad_structure(Offender, P,{?DOP_DEMONITOR_P,'replace',P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + start_monitor(Offender,P), send_bad_structure(Offender, P,{?DOP_DEMONITOR_P,'replace',P,normal},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), + send_bad_structure(Offender, P,{?DOP_EXIT,'replace',P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT,make_ref(),normal,normal},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT_TT,'replace',token,P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT_TT,make_ref(),token,normal,normal},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT2,'replace',P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT2,make_ref(),normal,normal},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT2_TT,'replace',token,P},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_EXIT2_TT,make_ref(),token,normal,normal},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_GROUP_LEADER,'replace'},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_GROUP_LEADER,'replace','atomic'},2), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_GROUP_LEADER,'replace',P},0), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND_TT,'replace','',name},2,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND_TT,'replace','',name,token},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND,'replace',''},2,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND,'replace','',P},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND,'replace','',name},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_REG_SEND,'replace','',name,{token}},2,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_SEND_TT,'',P},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_SEND_TT,'',name,token},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_SEND,''},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_SEND,'',name},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), send_bad_structure(Offender, P,{?DOP_SEND,'',P,{token}},0,{message}), - pong = rpc:call(Victim, net_adm, ping, [Offender]), P ! two, P ! check_msgs, receive @@ -1685,13 +1663,16 @@ bad_dist_ext_size(Config) when is_list(Config) -> start_node_monitors([Offender,Victim]), Parent = self(), - P = spawn_link(Victim, + P = spawn_opt(Victim, fun () -> Parent ! {self(), started}, receive check_msgs -> ok end, %% DID CRASH HERE bad_dist_ext_check_msgs([one]), Parent ! {self(), messages_checked} - end), + end, + [link, + %% on_heap to force total_heap_size to inspect msg queue + {message_queue_data, on_heap}]), receive {P, started} -> ok end, P ! one, @@ -1714,6 +1695,7 @@ bad_dist_ext_size(Config) when is_list(Config) -> verify_still_up(Offender, Victim), + %% Let process_info(P, total_heap_size) find bad msg and disconnect rpc:call(Victim, erlang, process_info, [P, total_heap_size]), verify_down(Offender, connection_closed, Victim, killed), @@ -1795,10 +1777,11 @@ send_bad_structure(Offender,Victim,Bad,WhereToPutSelf) -> send_bad_structure(Offender,Victim,Bad,WhereToPutSelf,PayLoad) -> Parent = self(), Done = make_ref(), - spawn(Offender, + spawn_link(Offender, fun () -> Node = node(Victim), pong = net_adm:ping(Node), + erlang:monitor_node(Node, true), DCtrl = dctrl(Node), Bad1 = case WhereToPutSelf of 0 -> @@ -1812,7 +1795,16 @@ send_bad_structure(Offender,Victim,Bad,WhereToPutSelf,PayLoad) -> [] -> []; _Other -> [dmsg_ext(PayLoad)] end, + + receive {nodedown, Node} -> exit("premature nodedown") + after 10 -> ok + end, + dctrl_send(DCtrl, DData), + + receive {nodedown, Node} -> ok + after 5000 -> exit("missing nodedown") + end, Parent ! {DData,Done} end), receive diff --git a/erts/emulator/test/process_SUITE.erl b/erts/emulator/test/process_SUITE.erl index a9f20f9928..a8bcfac84d 100644 --- a/erts/emulator/test/process_SUITE.erl +++ b/erts/emulator/test/process_SUITE.erl @@ -2532,8 +2532,13 @@ system_task_on_suspended(Config) when is_list(Config) -> end. gc_request_when_gc_disabled(Config) when is_list(Config) -> - Master = self(), AIS = erts_debug:set_internal_state(available_internal_state, true), + gc_request_when_gc_disabled_do(ref), + gc_request_when_gc_disabled_do(immed), + erts_debug:set_internal_state(available_internal_state, AIS). + +gc_request_when_gc_disabled_do(ReqIdType) -> + Master = self(), {P, M} = spawn_opt(fun () -> true = erts_debug:set_internal_state(gc_state, false), @@ -2545,7 +2550,10 @@ gc_request_when_gc_disabled(Config) when is_list(Config) -> receive after 100 -> ok end end, [monitor, link]), receive {P, gc_state, false} -> ok end, - ReqId = make_ref(), + ReqId = case ReqIdType of + ref -> make_ref(); + immed -> immed + end, async = garbage_collect(P, [{async, ReqId}]), receive {garbage_collect, ReqId, Result} -> @@ -2554,7 +2562,6 @@ gc_request_when_gc_disabled(Config) when is_list(Config) -> ok end, receive {garbage_collect, ReqId, true} -> ok end, - erts_debug:set_internal_state(available_internal_state, AIS), receive {'DOWN', M, process, P, _Reason} -> ok end, ok. diff --git a/erts/emulator/utils/beam_makeops b/erts/emulator/utils/beam_makeops index d7791d23fa..da994fae3e 100755 --- a/erts/emulator/utils/beam_makeops +++ b/erts/emulator/utils/beam_makeops @@ -19,7 +19,7 @@ # %CopyrightEnd% # use strict; -use vars qw($BEAM_FORMAT_NUMBER); +use vars qw($BEAM_FORMAT_NUMBER $GC_REGEXP); use constant COLD => 0; use constant WARM => 1; use constant HOT => 2; @@ -36,6 +36,7 @@ use constant PACK_CMD_LOOSE => '3'; use constant PACK_CMD_WIDE => '4'; $BEAM_FORMAT_NUMBER = undef; +$GC_REGEXP = undef; my $target = \&emulator_output; my $outdir = "."; # Directory for output files. @@ -77,6 +78,10 @@ my %num_specific; my %gen_to_spec; my %specific_op; +# The following hashes are used for error checking. +my %print_name; +my %specific_op_arity; + # Information about each specific operator. Key is the print name (e.g. get_list_xxy). # Value is a hash. my %spec_op_info; @@ -131,7 +136,10 @@ my $loader_types = "nprvlqo"; my $genop_types = $compiler_types . $loader_types; # -# Defines the argument types and their loaded size assuming no packing. +# Define the operand types and their loaded size assuming no packing. +# +# Those are the types that can be used in the definition of a specific +# instruction. # my %arg_size = ('r' => 0, # x(0) - x register zero 'x' => 1, # x(N), N > 0 - x register @@ -154,12 +162,35 @@ my %arg_size = ('r' => 0, # x(0) - x register zero 'A' => 1, # arity value 'P' => 1, # byte offset into tuple or stack 'Q' => 1, # like 'P', but packable - 'h' => 1, # character + 'h' => 1, # character (not used) 'l' => 1, # float reg 'q' => 1, # literal term ); # +# Define the types that may be used in a transformation rule. +# +# %pattern_type defines the types that may be used in a pattern +# on the left side. +# +# %construction_type defines the types that may be used when +# constructing a new instruction on the right side (a subset of +# the pattern types that are possible to construct). +# +my $pattern_types = "acdfjilnopqsuxy"; +my %pattern_type; +@pattern_type{split("", $pattern_types)} = (1) x length($pattern_types); + +my %construction_type; +foreach my $type (keys %pattern_type) { + $construction_type{$type} = 1 + if index($genop_types, $type) >= 0; +} +foreach my $makes_no_sense ('f', 'j', 'o', 'p', 'q') { + delete $construction_type{$makes_no_sense}; +} + +# # Generate bits. # my %type_bit; @@ -194,7 +225,8 @@ sub define_type_bit { define_type_bit('S', $type_bit{'d'}); define_type_bit('j', $type_bit{'f'} | $type_bit{'p'}); - # Aliases (for matching purposes). + # Aliases of 'u'. Those specify how to load the operand and + # what kind of packing can be done. define_type_bit('t', $type_bit{'u'}); define_type_bit('I', $type_bit{'u'}); define_type_bit('W', $type_bit{'u'}); @@ -279,9 +311,15 @@ if ($wordsize == 64) { # Add placeholders for built-in macros. # -$c_code{'IS_PACKED'} = ['$Expr',"built-in macro",('Expr')]; -$c_code{'ARG_POSITION'} = ['$Expr',"built-in macro",('Expr')]; -foreach my $name (keys %c_code) { +my %predef_macros = + (OPERAND_POSITION => ['Expr'], + IF => ['Expr','IfTrue','IfFalse'], + REFRESH_GEN_DEST => [], + ); +foreach my $name (keys %predef_macros) { + my @args = @{$predef_macros{$name}}; + my $body = join(':', map { '$' . $_ } @args); + $c_code{$name} = [$body,"built-in macro",@args], $c_code_used{$name} = 1; } @@ -359,8 +397,10 @@ while (<>) { # if (/^([\w_][\w\d_]+)=(.*)/) { no strict 'refs'; - my($name) = $1; - $$name = $2; + my $name = $1; + my $value = $2; + $value =~ s/;\s*$//; + $$name = $value; next; } @@ -1019,6 +1059,22 @@ sub parse_specific_op { my $key = "$name/$arity"; foreach my $args_ref (@res) { @args = @$args_ref; + my $arity = @args; + my $loc = "$ARGV($.)"; + if (defined $specific_op_arity{$name}) { + my($prev_arity,$loc) = @{$specific_op_arity{$name}}; + if ($arity != $prev_arity) { + error("$name defined with arity $arity, " . + "but previously defined with arity $prev_arity at $loc"); + } + } + $specific_op_arity{$name} = [$arity,$loc]; + my $print_name = print_name($name, @args); + if (defined $print_name{$print_name}) { + error("$name @args: already defined at " . + $print_name{$print_name}); + } + $print_name{$print_name} = $loc; push @{$specific_op{$key}}, [$name,$hotness,@args]; } @@ -1333,7 +1389,9 @@ sub cg_basic { # sub cg_combined_size { - my %params = (@_, pack_options => \@basic_pack_options); + my %params = (@_, + pack_options => \@basic_pack_options, + size_only => 1); $params{pack_options} = \@extended_pack_options if $params{first}; my($size) = code_gen(%params); @@ -1361,6 +1419,7 @@ sub code_gen { my %params = (extra_comments => '', offset => 0, inc => 0, + size_only => 0, @_); my $name = $params{name}; my $extra_comments = $params{extra_comments}; @@ -1393,6 +1452,7 @@ sub code_gen { my $need_block = 0; my $arg_offset = $offset; + my $has_gen_dest = 0; @args = map { s/[?]$//g; $_ } @args; foreach (@args) { my($this_size) = $arg_size{$_}; @@ -1403,6 +1463,7 @@ sub code_gen { "Eterm* dst_ptr = REG_TARGET_PTR(dst);\n"; push(@f, "*dst_ptr"); $this_size = $1; + $has_gen_dest = 1; last SWITCH; }; /^packed:[a-zA-z]:(\d):(.*)/ and do { @@ -1435,6 +1496,7 @@ sub code_gen { $var_decls .= "Eterm dst = " . arg_offset($arg_offset) . ";\n" . "Eterm* dst_ptr = REG_TARGET_PTR(dst);\n"; push(@f, "*dst_ptr"); + $has_gen_dest = 1; last SWITCH; }; defined $arg_size{$_} and do { @@ -1449,10 +1511,10 @@ sub code_gen { } # - # If the implementation is in beam_emu.c, there is nothing - # more to do. + # If the implementation is in beam_emu.c or if + # the caller only wants the size, we are done. # - unless (defined $c_code_ref) { + if (not defined $c_code_ref or $params{size_only}) { return ($size+1, undef, ''); } @@ -1517,9 +1579,36 @@ sub code_gen { "{", "$var_decls$body", "}", ""); + + # Make sure that $REFRESH_GEN_DEST() is used when a + # general destination ('d') may have been clobbered by + # a GC. + my $gc_error = verify_gc_code($code, $has_gen_dest); + if (defined $gc_error) { + warn $gc_error; + error("... from the body of $name at $where"); + } + + # Done. ($size+1, $code, $pack_spec); } +sub verify_gc_code { + my $code = shift; + my $has_gen_dest = shift; + + return unless $has_gen_dest; + + if ($code =~ /$GC_REGEXP/o) { + my $code_after_gc = substr($code, $+[0]); + unless ($code_after_gc =~ /dst_ptr = REG_TARGET_PTR/) { + return "pointer to destination register is invalid after GC -- " . + "use \$REFRESH_GEN_DEST()\n"; + } + } + return undef; +} + sub arg_offset { my $offset = shift; "I[" . ($offset+1) . "]"; @@ -1619,17 +1708,26 @@ sub expand_macro { } # Handle built-in macros. - if ($name eq 'ARG_POSITION') { + if ($name eq 'OPERAND_POSITION') { if ($body =~ /^I\[(\d+)\]$/) { $body = $1; } else { $body = 0; } - } elsif ($name eq 'IS_PACKED') { - $body = ($body =~ /^I\[\d+\]$/) ? 0 : 1; + } elsif ($name eq 'IF') { + my $expr = $new_bindings{Expr}; + my $bool = eval $expr; + if ($@ ne '') { + &error("bad expression '$expr' in \$IF()"); + } + my $part = $bool ? 'IfTrue' : 'IfFalse'; + $body = $new_bindings{$part}; + } elsif ($name eq 'REFRESH_GEN_DEST') { + $body = "dst_ptr = REG_TARGET_PTR(dst)"; } - # Wrap body if needed and return resul.t + + # Wrap body if needed and return result. $body = "do {\n$body\n} while (0)" if needs_do_wrapper($body); ($body,$rest); @@ -2156,12 +2254,19 @@ sub tr_parse_op { if (/^([a-z*]+)(.*)/) { $type = $1; $_ = $2; + error("$type: only a single type is allowed on right side of transformations") + if not $src and length($type) > 1; foreach (split('', $type)) { - error("bad type in $op") - unless defined $type_bit{$_} or $type eq '*'; - $_ eq 'r' and - error("$op: 'r' is not allowed in transformations") - } + next if $src and $type eq '*'; + error("$op: not a type") + unless defined $type_bit{$_}; + error("$op: the type '$_' is not allowed in transformations") + unless defined $pattern_type{$_}; + if (not $src) { + error("$op: type '$_' is not allowed on the right side of transformations") + unless defined $construction_type{$_}; + } + } } # Get an optional condition. (In source.) @@ -2194,10 +2299,18 @@ sub tr_parse_op { } # Get an optional value. (In destination.) - $type_val = $type eq 'x' ? 1023 : 0; + if ($type eq 'x') { + $type_val = 1023; + } elsif ($type eq 'a') { + $type_val = 'am_Empty'; + } else { + $type_val = 0; + } if (/^=(.*)/) { - error("value not allowed in source: $op") + error("$op: value not allowed in source") if $src; + error("$op: the type 'n' must not be given a value") + if $type eq 'n'; $type_val = $1; $_ = ''; } @@ -2207,13 +2320,16 @@ sub tr_parse_op { error("garbage '$_' after operand: $op") unless /^\s*$/; - # Test that destination has no conditions. + # Check the conditions. - unless ($src) { - error("condition not allowed in destination: $op") + if ($src) { + error("$op: the type '$type' is not allowed to be compared with a literal value") + if $cond and not $construction_type{$type}; + } else { + error("$op: condition not allowed in destination") if $cond; - error("variable name and type cannot be combined in destination: $op") - if $var && $type; + error("$op: variable name and type cannot be combined in destination") + if $var and $type; } ($var,$type,$type_val,$cond,$cond_val); diff --git a/erts/preloaded/ebin/zlib.beam b/erts/preloaded/ebin/zlib.beam Binary files differindex f388bc723a..4ad5f37434 100644 --- a/erts/preloaded/ebin/zlib.beam +++ b/erts/preloaded/ebin/zlib.beam diff --git a/erts/preloaded/src/zlib.erl b/erts/preloaded/src/zlib.erl index 03c9ae38a1..a4ef42204d 100644 --- a/erts/preloaded/src/zlib.erl +++ b/erts/preloaded/src/zlib.erl @@ -188,14 +188,13 @@ deflateReset_nif(_Z) -> deflateParams(Z, Level0, Strategy0) -> Level = arg_level(Level0), Strategy = arg_strategy(Strategy0), + Progress = deflate(Z, <<>>, sync), case deflateParams_nif(Z, Level, Strategy) of - buf_error -> - %% We had data left in the pipe; flush everything and stash it away - %% for the next deflate call before trying again. - Output = deflate(Z, <<>>, full), - save_progress(Z, deflate, Output), - deflateParams_nif(Z, Level, Strategy); - Any -> Any + ok -> + save_progress(Z, deflate, Progress), + ok; + Other -> + Other end. deflateParams_nif(_Z, _Level, _Strategy) -> erlang:nif_error(undef). diff --git a/erts/vsn.mk b/erts/vsn.mk index a788b2e491..220c3b5f6c 100644 --- a/erts/vsn.mk +++ b/erts/vsn.mk @@ -18,7 +18,7 @@ # %CopyrightEnd% # -VSN = 9.1.3 +VSN = 9.1.4 # Port number 4365 in 4.2 # Port number 4366 in 4.3 diff --git a/lib/common_test/src/cth_log_redirect.erl b/lib/common_test/src/cth_log_redirect.erl index 8b29d0f96d..1c55e17686 100644 --- a/lib/common_test/src/cth_log_redirect.erl +++ b/lib/common_test/src/cth_log_redirect.erl @@ -121,8 +121,8 @@ handle_event({_Type,GL,_Msg}, #eh_state{handle_remote_events = false} = State) when node(GL) /= node() -> {ok, State}; handle_event(Event, #eh_state{log_func = LogFunc} = State) -> - case lists:keyfind(sasl, 1, application:which_applications()) of - false -> + case whereis(sasl_sup) of + undefined -> sasl_not_started; _Else -> {ok, ErrLogType} = application:get_env(sasl, errlog_type), diff --git a/lib/compiler/internal_doc/cerl-notes.md b/lib/compiler/internal_doc/cerl-notes.md new file mode 100644 index 0000000000..705fe8f42d --- /dev/null +++ b/lib/compiler/internal_doc/cerl-notes.md @@ -0,0 +1,75 @@ +Some notes on the cerl modules +============================== + +Maps in cerl_clauses:match/3 +---------------------------- + +Not much optimization is done for maps in `cerl_clauses:match/3`. + +The reason is that the inliner (`cerl_inline`) was not designed for +data types that depend on variables bound in the enclosing environment +(for example, the keys for maps). If we attempt to extend the +optimizations for maps similar to the optimizations for the other data +types, the inliner will crash for certain code. Here is an example of +code that would crash the inliner: + + t() -> + f(key). + + f(K) -> + case #{K => value} of + #{key := V} -> V + end. + +The reason for the crash is that the inliner works in several +passes and calls `cerl_clauses:match/3` twice. The inliner will +assume that the same result will be returned both times, but +for this example different bindings will be returned. + +Here is the rough outline of what happens: + +* The first time `cerl_clauses:match/3` will be asked to match the +pattern `#{K := V}` against `#{key => value}`. It cannot say more +than that the pattern *may* match. + +* Now the inliner will add the bindings to body of the clause (which +is simply `V`). In this case, the bindings are still empty, so +nothing is added. + +* The inliner will now do some substitutions and renaming. The +variable `K` will be replaced with `key`. + +* The next time `cerl_clauses:match/3` is called, it will be asked to +match the pattern `#{key := V}` against `#{key => value#}`. In this +case, there will be a match and the bindings can be extended with +`{V,value}`. + +* The inliner will see that the entire case can be removed. Only +the body for the clause needs to be kept. + +Thus, after inlining the function `t/0` will look like this: + + t() -> + V. + +The problem here is that the inliner assumed that the bindings from +the first and second call to `cer_clauses:match/3` would be the same. +It used the empty bindings from the first call for the body. + +The correct way would be to add the bindings from the second call: + + t() -> + let V = value in V. + +### How to fix this ### + +The inliner will need to call `cerl_clauses:match/3` after doing +all substitutions and renaming. It is not clear to me how difficult +that would be. I assume that the inliner is written the way it is +for a good reason. That means that switching the order things are +done would lead to correctness and/or performance problems. + +### What must also be done to fix this ### + +`cerl_inline:make_template/3` must be extended to create a template +for maps. That is relatively straightforward. diff --git a/lib/compiler/src/cerl_clauses.erl b/lib/compiler/src/cerl_clauses.erl index 7d6518c3c6..fa5104c01b 100644 --- a/lib/compiler/src/cerl_clauses.erl +++ b/lib/compiler/src/cerl_clauses.erl @@ -353,6 +353,8 @@ match(P, E, Bs) -> map -> %% The most we can do is to say "definitely no match" if a %% map pattern is matched against non-map data. + %% (Note: See the document internal_doc/cerl-notes.md for + %% information why we don't try to do more here.) case E of any -> {false, Bs}; diff --git a/lib/compiler/src/compile.erl b/lib/compiler/src/compile.erl index bc519264fc..b0d0da0f66 100644 --- a/lib/compiler/src/compile.erl +++ b/lib/compiler/src/compile.erl @@ -1448,16 +1448,16 @@ beam_asm(Code0, #compile{ifile=File,extra_chunks=ExtraChunks,options=CompilerOpt {ok,DebugInfo,Opts0} -> Opts1 = [O || O <- Opts0, effects_code_generation(O)], Chunks = [{<<"Dbgi">>, DebugInfo} | ExtraChunks], - CompileInfo = compile_info(File, Opts1), + CompileInfo = compile_info(File, CompilerOpts, Opts1), {ok,Code} = beam_asm:module(Code0, Chunks, CompileInfo, CompilerOpts), {ok,Code,St#compile{abstract_code=[]}}; {error,Es} -> {error,St#compile{errors=St#compile.errors ++ [{File,Es}]}} end. -compile_info(File, Opts) -> - IsSlim = member(slim, Opts), - IsDeterministic = member(deterministic, Opts), +compile_info(File, CompilerOpts, Opts) -> + IsSlim = member(slim, CompilerOpts), + IsDeterministic = member(deterministic, CompilerOpts), Info0 = proplists:get_value(compile_info, Opts, []), Info1 = case paranoid_absname(File) of diff --git a/lib/compiler/test/compile_SUITE.erl b/lib/compiler/test/compile_SUITE.erl index e88a132d59..8fe2a93f95 100644 --- a/lib/compiler/test/compile_SUITE.erl +++ b/lib/compiler/test/compile_SUITE.erl @@ -120,9 +120,19 @@ file_1(Config) when is_list(Config) -> true = exists(Target), passed = run(Target, test, []), + %% Test option 'deterministic' as a compiler attribute. + Det = deterministic_module, + {DetPath, DetTarget} = get_files(Config, Det, "det_target"), + {ok,Det,DetCode} = compile:file(DetPath, [binary]), + {module,Det} = code:load_binary(Det, "", DetCode), + [{version,_}] = Det:module_info(compile), + true = code:delete(Det), + false = code:purge(Det), + %% Cleanup. ok = file:delete(Target), ok = file:del_dir(filename:dirname(Target)), + ok = file:del_dir(filename:dirname(DetTarget)), %% There should not be any messages in the messages. receive @@ -1318,10 +1328,13 @@ do_warnings_2([], Next, F) -> %% pre-loads the modules that are used by a typical compilation. pre_load_check(Config) -> - case test_server:is_cover() of - true -> + case {test_server:is_cover(),code:module_info(native)} of + {true,_} -> {skip,"Cover is running"}; - false -> + {false,true} -> + %% Tracing won't work. + {skip,"'code' is native-compiled"}; + {false,false} -> try do_pre_load_check(Config) after diff --git a/lib/compiler/test/compile_SUITE_data/deterministic_module.erl b/lib/compiler/test/compile_SUITE_data/deterministic_module.erl new file mode 100644 index 0000000000..5e0e29c25e --- /dev/null +++ b/lib/compiler/test/compile_SUITE_data/deterministic_module.erl @@ -0,0 +1,21 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +-module(deterministic_module). +-compile([deterministic]). diff --git a/lib/crypto/c_src/Makefile.in b/lib/crypto/c_src/Makefile.in index af7c209c75..31124ba477 100644 --- a/lib/crypto/c_src/Makefile.in +++ b/lib/crypto/c_src/Makefile.in @@ -1,7 +1,7 @@ # # %CopyrightBegin% # -# Copyright Ericsson AB 1999-2016. All Rights Reserved. +# Copyright Ericsson AB 1999-2017. All Rights Reserved. # # Licensed under the Apache License, Version 2.0 (the "License"); # you may not use this file except in compliance with the License. @@ -78,12 +78,16 @@ CRYPTO_STATIC_OBJS = $(OBJDIR)/crypto_static$(TYPEMARKER).o\ NIF_ARCHIVE = $(LIBDIR)/crypto$(TYPEMARKER).a +TEST_ENGINE_OBJS = $(OBJDIR)/otp_test_engine$(TYPEMARKER).o + ifeq ($(findstring win32,$(TARGET)), win32) NIF_LIB = $(LIBDIR)/crypto$(TYPEMARKER).dll CALLBACK_LIB = $(LIBDIR)/crypto_callback$(TYPEMARKER).dll +TEST_ENGINE_LIB = $(LIBDIR)/otp_test_engine$(TYPEMARKER).dll else NIF_LIB = $(LIBDIR)/crypto$(TYPEMARKER).so CALLBACK_LIB = $(LIBDIR)/crypto_callback$(TYPEMARKER).so +TEST_ENGINE_LIB = $(LIBDIR)/otp_test_engine$(TYPEMARKER).so endif ifeq ($(HOST_OS),) @@ -129,10 +133,22 @@ ALL_STATIC_CFLAGS = $(DED_STATIC_CFLAGS) $(INCLUDES) _create_dirs := $(shell mkdir -p $(OBJDIR) $(LIBDIR)) -debug opt valgrind: $(NIF_LIB) $(CALLBACK_LIB) +debug opt valgrind: $(NIF_LIB) $(CALLBACK_LIB) $(TEST_ENGINE_LIB) static_lib: $(NIF_ARCHIVE) +$(OBJDIR)/otp_test_engine$(TYPEMARKER).o: otp_test_engine.c + $(V_at)$(INSTALL_DIR) $(OBJDIR) + $(V_CC) -c -o $@ $(ALL_CFLAGS) $< + +$(LIBDIR)/otp_test_engine$(TYPEMARKER).so: $(TEST_ENGINE_OBJS) + $(V_at)$(INSTALL_DIR) $(LIBDIR) + $(V_LD) $(LDFLAGS) -o $@ $^ $(LDLIBS) $(CRYPTO_LINK_LIB) + +$(LIBDIR)/otp_test_engine$(TYPEMARKER).dll: $(TEST_ENGINE_OBJS) + $(V_at)$(INSTALL_DIR) $(LIBDIR) + $(V_LD) $(LDFLAGS) -o $@ $(SSL_DED_LD_RUNTIME_LIBRARY_PATH) -L$(SSL_LIBDIR) $(TEST_ENGINE_OBJS) -l$(SSL_CRYPTO_LIBNAME) -l$(SSL_SSL_LIBNAME) + $(OBJDIR)/%$(TYPEMARKER).o: %.c $(V_at)$(INSTALL_DIR) $(OBJDIR) $(V_CC) -c -o $@ $(ALL_CFLAGS) $< @@ -170,6 +186,7 @@ ifeq ($(findstring win32,$(TARGET)), win32) rm -f $(LIBDIR)/crypto.debug.dll rm -f $(LIBDIR)/crypto_callback.dll rm -f $(LIBDIR)/crypto_callback.debug.dll + rm -f $(LIBDIR)/otp_test_engine.dll else rm -f $(LIBDIR)/crypto.so rm -f $(LIBDIR)/crypto.debug.so @@ -177,6 +194,7 @@ else rm -f $(LIBDIR)/crypto_callback.so rm -f $(LIBDIR)/crypto_callback.debug.so rm -f $(LIBDIR)/crypto_callback.valgrind.so + rm -f $(LIBDIR)/otp_test_engine.so endif rm -f $(OBJDIR)/crypto.o rm -f $(OBJDIR)/crypto_static.o @@ -187,6 +205,7 @@ endif rm -f $(OBJDIR)/crypto_callback.o rm -f $(OBJDIR)/crypto_callback.debug.o rm -f $(OBJDIR)/crypto_callback.valgrind.o + rm -f $(OBJDIR)/otp_test_engine.o rm -f core *~ docs: @@ -206,6 +225,8 @@ ifeq ($(DYNAMIC_CRYPTO_LIB),yes) $(INSTALL_PROGRAM) $(CALLBACK_OBJS) "$(RELSYSDIR)/priv/obj" $(INSTALL_PROGRAM) $(CALLBACK_LIB) "$(RELSYSDIR)/priv/lib" endif + $(INSTALL_PROGRAM) $(TEST_ENGINE_OBJS) "$(RELSYSDIR)/priv/obj" + $(INSTALL_PROGRAM) $(TEST_ENGINE_LIB) "$(RELSYSDIR)/priv/lib" release_docs_spec: diff --git a/lib/crypto/c_src/crypto.c b/lib/crypto/c_src/crypto.c index c4e80e3153..b29c5082ba 100644 --- a/lib/crypto/c_src/crypto.c +++ b/lib/crypto/c_src/crypto.c @@ -1,4 +1,4 @@ -/* +/* * %CopyrightBegin% * * Copyright Ericsson AB 2010-2017. All Rights Reserved. @@ -19,8 +19,8 @@ */ /* - * Purpose: Dynamically loadable NIF library for cryptography. - * Based on OpenSSL. + * Purpose: Dynamically loadable NIF library for cryptography. + * Based on OpenSSL. */ #ifdef __WIN32__ @@ -60,6 +60,8 @@ #include <openssl/rand.h> #include <openssl/evp.h> #include <openssl/hmac.h> +#include <openssl/engine.h> +#include <openssl/err.h> /* Helper macro to construct a OPENSSL_VERSION_NUMBER. * See openssl/opensslv.h @@ -79,9 +81,9 @@ * * Therefor works tests like this as intendend: * OPENSSL_VERSION_NUMBER >= PACKED_OPENSSL_VERSION_PLAIN(1,0,0) - * (The test is for example "2.4.2" >= "1.0.0" although the test + * (The test is for example "2.4.2" >= "1.0.0" although the test * with the cloned OpenSSL test would be "1.0.1" >= "1.0.0") - * + * * But tests like this gives wrong result: * OPENSSL_VERSION_NUMBER < PACKED_OPENSSL_VERSION_PLAIN(1,1,0) * (The test is false since "2.4.2" < "1.1.0". It should have been @@ -119,6 +121,10 @@ #include <openssl/modes.h> #endif +#if OPENSSL_VERSION_NUMBER >= PACKED_OPENSSL_VERSION(0,9,8,'h') +#define HAS_ENGINE_SUPPORT +#endif + #include "crypto_callback.h" #if OPENSSL_VERSION_NUMBER >= PACKED_OPENSSL_VERSION_PLAIN(0,9,8) \ @@ -240,7 +246,7 @@ /* This shall correspond to the similar macro in crypto.erl */ /* Current value is: erlang:system_info(context_reductions) * 10 */ -#define MAX_BYTES_TO_NIF 20000 +#define MAX_BYTES_TO_NIF 20000 #define CONSUME_REDS(NifEnv, Ibin) \ do { \ @@ -342,6 +348,10 @@ static INLINE void RSA_get0_crt_params(const RSA *r, const BIGNUM **dmp1, const static INLINE int DSA_set0_key(DSA *d, BIGNUM *pub_key, BIGNUM *priv_key); static INLINE int DSA_set0_pqg(DSA *d, BIGNUM *p, BIGNUM *q, BIGNUM *g); +static INLINE void DSA_get0_pqg(const DSA *dsa, + const BIGNUM **p, const BIGNUM **q, const BIGNUM **g); +static INLINE void DSA_get0_key(const DSA *dsa, + const BIGNUM **pub_key, const BIGNUM **priv_key); static INLINE int DSA_set0_key(DSA *d, BIGNUM *pub_key, BIGNUM *priv_key) { @@ -358,6 +368,23 @@ static INLINE int DSA_set0_pqg(DSA *d, BIGNUM *p, BIGNUM *q, BIGNUM *g) return 1; } +static INLINE void +DSA_get0_pqg(const DSA *dsa, const BIGNUM **p, const BIGNUM **q, const BIGNUM **g) +{ + *p = dsa->p; + *q = dsa->q; + *g = dsa->g; +} + +static INLINE void +DSA_get0_key(const DSA *dsa, const BIGNUM **pub_key, const BIGNUM **priv_key) +{ + if (pub_key) *pub_key = dsa->pub_key; + if (priv_key) *priv_key = dsa->priv_key; +} + + + static INLINE int DH_set0_key(DH *dh, BIGNUM *pub_key, BIGNUM *priv_key); static INLINE int DH_set0_pqg(DH *dh, BIGNUM *p, BIGNUM *q, BIGNUM *g); static INLINE int DH_set_length(DH *dh, long length); @@ -387,6 +414,8 @@ static INLINE int DH_set_length(DH *dh, long length) return 1; } + + static INLINE void DH_get0_pqg(const DH *dh, const BIGNUM **p, const BIGNUM **q, const BIGNUM **g) { @@ -398,8 +427,8 @@ DH_get0_pqg(const DH *dh, const BIGNUM **p, const BIGNUM **q, const BIGNUM **g) static INLINE void DH_get0_key(const DH *dh, const BIGNUM **pub_key, const BIGNUM **priv_key) { - *pub_key = dh->pub_key; - *priv_key = dh->priv_key; + if (pub_key) *pub_key = dh->pub_key; + if (priv_key) *priv_key = dh->priv_key; } #else /* End of compatibility definitions. */ @@ -448,6 +477,7 @@ static ERL_NIF_TERM dh_generate_parameters_nif(ErlNifEnv* env, int argc, const E static ERL_NIF_TERM dh_check(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM dh_generate_key_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM dh_compute_key_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM privkey_to_pubkey_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM srp_value_B_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM srp_user_secret_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM srp_host_secret_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); @@ -466,6 +496,22 @@ static ERL_NIF_TERM aes_gcm_decrypt_NO_EVP(ErlNifEnv* env, int argc, const ERL_N static ERL_NIF_TERM chacha20_poly1305_encrypt(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); static ERL_NIF_TERM chacha20_poly1305_decrypt(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static int get_engine_load_cmd_list(ErlNifEnv* env, const ERL_NIF_TERM term, char **cmds, int i); +static ERL_NIF_TERM engine_by_id_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_init_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_finish_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_free_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_load_dynamic_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_ctrl_cmd_strings_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_register_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_unregister_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_add_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_remove_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_get_first_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_get_next_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_get_id_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); +static ERL_NIF_TERM engine_get_all_methods_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]); + /* helpers */ static void init_algorithms_types(ErlNifEnv*); static void init_digest_types(ErlNifEnv* env); @@ -477,6 +523,10 @@ static int term2point(ErlNifEnv* env, ERL_NIF_TERM term, #endif static ERL_NIF_TERM bin_from_bn(ErlNifEnv* env, const BIGNUM *bn); +#ifdef HAS_ENGINE_SUPPORT +static int zero_terminate(ErlNifBinary bin, char **buf); +#endif + static int library_refc = 0; /* number of users of this dynamic library */ static ErlNifFunc nif_funcs[] = { @@ -516,6 +566,7 @@ static ErlNifFunc nif_funcs[] = { {"dh_check", 1, dh_check}, {"dh_generate_key_nif", 4, dh_generate_key_nif}, {"dh_compute_key_nif", 3, dh_compute_key_nif}, + {"privkey_to_pubkey_nif", 2, privkey_to_pubkey_nif}, {"srp_value_B_nif", 5, srp_value_B_nif}, {"srp_user_secret_nif", 7, srp_user_secret_nif}, {"srp_host_secret_nif", 5, srp_host_secret_nif}, @@ -529,12 +580,27 @@ static ErlNifFunc nif_funcs[] = { {"aes_gcm_decrypt", 5, aes_gcm_decrypt}, {"chacha20_poly1305_encrypt", 4, chacha20_poly1305_encrypt}, - {"chacha20_poly1305_decrypt", 5, chacha20_poly1305_decrypt} + {"chacha20_poly1305_decrypt", 5, chacha20_poly1305_decrypt}, + + {"engine_by_id_nif", 1, engine_by_id_nif}, + {"engine_init_nif", 1, engine_init_nif}, + {"engine_finish_nif", 1, engine_finish_nif}, + {"engine_free_nif", 1, engine_free_nif}, + {"engine_load_dynamic_nif", 0, engine_load_dynamic_nif}, + {"engine_ctrl_cmd_strings_nif", 2, engine_ctrl_cmd_strings_nif}, + {"engine_register_nif", 2, engine_register_nif}, + {"engine_unregister_nif", 2, engine_unregister_nif}, + {"engine_add_nif", 1, engine_add_nif}, + {"engine_remove_nif", 1, engine_remove_nif}, + {"engine_get_first_nif", 0, engine_get_first_nif}, + {"engine_get_next_nif", 1, engine_get_next_nif}, + {"engine_get_id_nif", 1, engine_get_id_nif}, + {"engine_get_all_methods_nif", 0, engine_get_all_methods_nif} + }; ERL_NIF_INIT(crypto,nif_funcs,load,NULL,upgrade,unload) - #define MD5_CTX_LEN (sizeof(MD5_CTX)) #define MD4_CTX_LEN (sizeof(MD4_CTX)) #define RIPEMD160_CTX_LEN (sizeof(RIPEMD160_CTX)) @@ -603,7 +669,33 @@ static ERL_NIF_TERM atom_sha512; static ERL_NIF_TERM atom_md5; static ERL_NIF_TERM atom_ripemd160; - +#ifdef HAS_ENGINE_SUPPORT +static ERL_NIF_TERM atom_bad_engine_method; +static ERL_NIF_TERM atom_bad_engine_id; +static ERL_NIF_TERM atom_ctrl_cmd_failed; +static ERL_NIF_TERM atom_engine_init_failed; +static ERL_NIF_TERM atom_register_engine_failed; +static ERL_NIF_TERM atom_add_engine_failed; +static ERL_NIF_TERM atom_remove_engine_failed; +static ERL_NIF_TERM atom_engine_method_not_supported; + +static ERL_NIF_TERM atom_engine_method_rsa; +static ERL_NIF_TERM atom_engine_method_dsa; +static ERL_NIF_TERM atom_engine_method_dh; +static ERL_NIF_TERM atom_engine_method_rand; +static ERL_NIF_TERM atom_engine_method_ecdh; +static ERL_NIF_TERM atom_engine_method_ecdsa; +static ERL_NIF_TERM atom_engine_method_ciphers; +static ERL_NIF_TERM atom_engine_method_digests; +static ERL_NIF_TERM atom_engine_method_store; +static ERL_NIF_TERM atom_engine_method_pkey_meths; +static ERL_NIF_TERM atom_engine_method_pkey_asn1_meths; +static ERL_NIF_TERM atom_engine_method_ec; + +static ERL_NIF_TERM atom_engine; +static ERL_NIF_TERM atom_key_id; +static ERL_NIF_TERM atom_password; +#endif static ErlNifResourceType* hmac_context_rtype; struct hmac_context @@ -728,11 +820,13 @@ static struct cipher_type_t cipher_types[] = static struct cipher_type_t* get_cipher_type(ERL_NIF_TERM type, size_t key_len); + /* #define PRINTF_ERR0(FMT) enif_fprintf(stderr, FMT "\n") #define PRINTF_ERR1(FMT, A1) enif_fprintf(stderr, FMT "\n", A1) #define PRINTF_ERR2(FMT, A1, A2) enif_fprintf(stderr, FMT "\n", A1, A2) */ + #define PRINTF_ERR0(FMT) #define PRINTF_ERR1(FMT,A1) #define PRINTF_ERR2(FMT,A1,A2) @@ -758,6 +852,23 @@ static void evp_cipher_ctx_dtor(ErlNifEnv* env, struct evp_cipher_ctx* ctx) { } #endif +// Engine +#ifdef HAS_ENGINE_SUPPORT +static ErlNifResourceType* engine_ctx_rtype; +struct engine_ctx { + ENGINE *engine; + char *id; +}; +static void engine_ctx_dtor(ErlNifEnv* env, struct engine_ctx* ctx) { + PRINTF_ERR0("engine_ctx_dtor"); + if(ctx->id) { + PRINTF_ERR1(" non empty ctx->id=%s", ctx->id); + enif_free(ctx->id); + } else + PRINTF_ERR0(" empty ctx->id=NULL"); +} +#endif + static int verify_lib_version(void) { const unsigned long libv = SSLeay(); @@ -793,7 +904,7 @@ static char crypto_callback_name[] = "crypto_callback"; static int change_basename(ErlNifBinary* bin, char* buf, int bufsz, const char* newfile) { int i; - + for (i = bin->size; i > 0; i--) { if (bin->data[i-1] == '/') break; @@ -869,12 +980,23 @@ static int initialize(ErlNifEnv* env, ERL_NIF_TERM load_info) return __LINE__; } #endif +#ifdef HAS_ENGINE_SUPPORT + engine_ctx_rtype = enif_open_resource_type(env, NULL, "ENGINE_CTX", + (ErlNifResourceDtor*) engine_ctx_dtor, + ERL_NIF_RT_CREATE|ERL_NIF_RT_TAKEOVER, + NULL); + if (!engine_ctx_rtype) { + PRINTF_ERR0("CRYPTO: Could not open resource type 'ENGINE_CTX'"); + return __LINE__; + } + if (library_refc > 0) { /* Repeated loading of this library (module upgrade). * Atoms and callbacks are already set, we are done. */ return 0; } +#endif atom_true = enif_make_atom(env,"true"); atom_false = enif_make_atom(env,"false"); @@ -952,6 +1074,33 @@ static int initialize(ErlNifEnv* env, ERL_NIF_TERM load_info) atom_md5 = enif_make_atom(env,"md5"); atom_ripemd160 = enif_make_atom(env,"ripemd160"); +#ifdef HAS_ENGINE_SUPPORT + atom_bad_engine_method = enif_make_atom(env,"bad_engine_method"); + atom_bad_engine_id = enif_make_atom(env,"bad_engine_id"); + atom_ctrl_cmd_failed = enif_make_atom(env,"ctrl_cmd_failed"); + atom_engine_init_failed = enif_make_atom(env,"engine_init_failed"); + atom_engine_method_not_supported = enif_make_atom(env,"engine_method_not_supported"); + atom_add_engine_failed = enif_make_atom(env,"add_engine_failed"); + atom_remove_engine_failed = enif_make_atom(env,"remove_engine_failed"); + + atom_engine_method_rsa = enif_make_atom(env,"engine_method_rsa"); + atom_engine_method_dsa = enif_make_atom(env,"engine_method_dsa"); + atom_engine_method_dh = enif_make_atom(env,"engine_method_dh"); + atom_engine_method_rand = enif_make_atom(env,"engine_method_rand"); + atom_engine_method_ecdh = enif_make_atom(env,"engine_method_ecdh"); + atom_engine_method_ecdsa = enif_make_atom(env,"engine_method_ecdsa"); + atom_engine_method_store = enif_make_atom(env,"engine_method_store"); + atom_engine_method_ciphers = enif_make_atom(env,"engine_method_ciphers"); + atom_engine_method_digests = enif_make_atom(env,"engine_method_digests"); + atom_engine_method_pkey_meths = enif_make_atom(env,"engine_method_pkey_meths"); + atom_engine_method_pkey_asn1_meths = enif_make_atom(env,"engine_method_pkey_asn1_meths"); + atom_engine_method_ec = enif_make_atom(env,"engine_method_ec"); + + atom_engine = enif_make_atom(env,"engine"); + atom_key_id = enif_make_atom(env,"key_id"); + atom_password = enif_make_atom(env,"password"); +#endif + init_digest_types(env); init_cipher_types(env); init_algorithms_types(env); @@ -973,24 +1122,24 @@ static int initialize(ErlNifEnv* env, ERL_NIF_TERM load_info) #else /* !HAVE_DYNAMIC_CRYPTO_LIB */ funcp = &get_crypto_callbacks; #endif - + #ifdef OPENSSL_THREADS enif_system_info(&sys_info, sizeof(sys_info)); if (sys_info.scheduler_threads > 1) { - nlocks = CRYPTO_num_locks(); + nlocks = CRYPTO_num_locks(); } /* else no need for locks */ #endif - + ccb = (*funcp)(nlocks); - + if (!ccb || ccb->sizeof_me != sizeof(*ccb)) { PRINTF_ERR0("Invalid 'crypto_callbacks'"); return __LINE__; } - + CRYPTO_set_mem_functions(ccb->crypto_alloc, ccb->crypto_realloc, ccb->crypto_free); - + #ifdef OPENSSL_THREADS if (nlocks > 0) { CRYPTO_set_locking_callback(ccb->locking_function); @@ -1186,11 +1335,11 @@ static ERL_NIF_TERM info_lib(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[] * Version string is still from library though. */ - memcpy(enif_make_new_binary(env, name_sz, &name_term), libname, name_sz); + memcpy(enif_make_new_binary(env, name_sz, &name_term), libname, name_sz); memcpy(enif_make_new_binary(env, ver_sz, &ver_term), ver, ver_sz); return enif_make_list1(env, enif_make_tuple3(env, name_term, - enif_make_int(env, ver_num), + enif_make_int(env, ver_num), ver_term)); } @@ -1225,6 +1374,8 @@ static ERL_NIF_TERM enable_fips_mode(ErlNifEnv* env, int argc, const ERL_NIF_TER } } + +#if defined(HAVE_EC) static ERL_NIF_TERM make_badarg_maybe(ErlNifEnv* env) { ERL_NIF_TERM reason; @@ -1233,6 +1384,7 @@ static ERL_NIF_TERM make_badarg_maybe(ErlNifEnv* env) else return enif_make_badarg(env); } +#endif static ERL_NIF_TERM hash_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) {/* (Type, Data) */ @@ -1668,7 +1820,7 @@ static ERL_NIF_TERM hmac_update_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM {/* (Context, Data) */ ErlNifBinary data; struct hmac_context* obj; - + if (!enif_get_resource(env, argv[0], hmac_context_rtype, (void**)&obj) || !enif_inspect_iolist_as_binary(env, argv[1], &data)) { return enif_make_badarg(env); @@ -1704,13 +1856,13 @@ static ERL_NIF_TERM hmac_final_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM enif_mutex_unlock(obj->mtx); return enif_make_badarg(env); } - + HMAC_Final(obj->ctx, mac_buf, &mac_len); HMAC_CTX_free(obj->ctx); obj->alive = 0; enif_mutex_unlock(obj->mtx); - if (argc == 2 && req_len < mac_len) { + if (argc == 2 && req_len < mac_len) { /* Only truncate to req_len bytes if asked. */ mac_len = req_len; } @@ -2021,7 +2173,7 @@ static ERL_NIF_TERM aes_ctr_stream_init(ErlNifEnv* env, int argc, const ERL_NIF_ } static ERL_NIF_TERM aes_ctr_stream_encrypt(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) -{/* ({Key, IVec, ECount, Num}, Data) */ +{/* ({Key, IVec, ECount, Num}, Data) */ ErlNifBinary key_bin, ivec_bin, text_bin, ecount_bin; AES_KEY aes_key; unsigned int num; @@ -2042,14 +2194,14 @@ static ERL_NIF_TERM aes_ctr_stream_encrypt(ErlNifEnv* env, int argc, const ERL_N return enif_make_badarg(env); } - ivec2_buf = enif_make_new_binary(env, ivec_bin.size, &ivec2_term); + ivec2_buf = enif_make_new_binary(env, ivec_bin.size, &ivec2_term); ecount2_buf = enif_make_new_binary(env, ecount_bin.size, &ecount2_term); - + memcpy(ivec2_buf, ivec_bin.data, 16); memcpy(ecount2_buf, ecount_bin.data, ecount_bin.size); AES_ctr128_encrypt((unsigned char *) text_bin.data, - enif_make_new_binary(env, text_bin.size, &cipher_term), + enif_make_new_binary(env, text_bin.size, &cipher_term), text_bin.size, &aes_key, ivec2_buf, ecount2_buf, &num); num2_term = enif_make_uint(env, num); @@ -2352,7 +2504,7 @@ out_err: } static ERL_NIF_TERM strong_rand_bytes_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) -{/* (Bytes) */ +{/* (Bytes) */ unsigned bytes; unsigned char* data; ERL_NIF_TERM ret; @@ -2446,7 +2598,7 @@ static ERL_NIF_TERM rand_uniform_nif(ErlNifEnv* env, int argc, const ERL_NIF_TER bn_to = BN_new(); BN_sub(bn_to, bn_rand, bn_from); - BN_pseudo_rand_range(bn_rand, bn_to); + BN_pseudo_rand_range(bn_rand, bn_to); BN_add(bn_rand, bn_rand, bn_from); dlen = BN_num_bytes(bn_rand); data = enif_make_new_binary(env, dlen+4, &ret); @@ -2464,7 +2616,7 @@ static ERL_NIF_TERM mod_exp_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM arg BIGNUM *bn_base=NULL, *bn_exponent=NULL, *bn_modulo=NULL, *bn_result; BN_CTX *bn_ctx; unsigned char* ptr; - unsigned dlen; + unsigned dlen; unsigned bin_hdr; /* return type: 0=plain binary, 4: mpint */ unsigned extra_byte; ERL_NIF_TERM ret; @@ -2485,7 +2637,7 @@ static ERL_NIF_TERM mod_exp_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM arg dlen = BN_num_bytes(bn_result); extra_byte = bin_hdr && BN_is_bit_set(bn_result, dlen*8-1); ptr = enif_make_new_binary(env, bin_hdr+extra_byte+dlen, &ret); - if (bin_hdr) { + if (bin_hdr) { put_int32(ptr, extra_byte+dlen); ptr[4] = 0; /* extra zeroed byte to ensure a positive mpint */ ptr += bin_hdr + extra_byte; @@ -2545,6 +2697,7 @@ static struct cipher_type_t* get_cipher_type(ERL_NIF_TERM type, size_t key_len) return NULL; } + static ERL_NIF_TERM do_exor(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) {/* (Data1, Data2) */ ErlNifBinary d1, d2; @@ -2578,7 +2731,7 @@ static ERL_NIF_TERM rc4_set_key(ErlNifEnv* env, int argc, const ERL_NIF_TERM arg return enif_make_badarg(env); } RC4_set_key((RC4_KEY*)enif_make_new_binary(env, sizeof(RC4_KEY), &ret), - key.size, key.data); + key.size, key.data); return ret; #else return enif_raise_exception(env, atom_notsup); @@ -2871,7 +3024,7 @@ static ERL_NIF_TERM dh_generate_parameters_nif(ErlNifEnv* env, int argc, const E BN_bn2bin(dh_g, g_ptr); ERL_VALGRIND_MAKE_MEM_DEFINED(p_ptr, p_len); ERL_VALGRIND_MAKE_MEM_DEFINED(g_ptr, g_len); - return enif_make_list2(env, ret_p, ret_g); + return enif_make_list2(env, ret_p, ret_g); } static ERL_NIF_TERM dh_check(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) @@ -2881,9 +3034,9 @@ static ERL_NIF_TERM dh_check(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[] ERL_NIF_TERM ret, head, tail; BIGNUM *dh_p, *dh_g; - if (!enif_get_list_cell(env, argv[0], &head, &tail) + if (!enif_get_list_cell(env, argv[0], &head, &tail) || !get_bn_from_bin(env, head, &dh_p) - || !enif_get_list_cell(env, tail, &head, &tail) + || !enif_get_list_cell(env, tail, &head, &tail) || !get_bn_from_bin(env, head, &dh_g) || !enif_is_empty_list(env,tail)) { @@ -2900,12 +3053,12 @@ static ERL_NIF_TERM dh_check(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[] else if (i & DH_NOT_SUITABLE_GENERATOR) ret = atom_not_suitable_generator; else ret = enif_make_tuple2(env, atom_unknown, enif_make_uint(env, i)); } - else { /* Check Failed */ + else { /* Check Failed */ ret = enif_make_tuple2(env, atom_error, atom_check_failed); } DH_free(dh_params); return ret; -} +} static ERL_NIF_TERM dh_generate_key_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) {/* (PrivKey|undefined, DHParams=[P,G], Mpint, Len|0) */ @@ -3007,7 +3160,7 @@ static ERL_NIF_TERM dh_compute_key_nif(ErlNifEnv* env, int argc, const ERL_NIF_T i = DH_compute_key(ret_bin.data, other_pub_key, dh_params); if (i > 0) { if (i != ret_bin.size) { - enif_realloc_binary(&ret_bin, i); + enif_realloc_binary(&ret_bin, i); } ret = enif_make_binary(env, &ret_bin); } @@ -3803,9 +3956,69 @@ static int get_pkey_sign_options(ErlNifEnv *env, ERL_NIF_TERM algorithm, ERL_NIF } +#ifdef HAS_ENGINE_SUPPORT +static int get_engine_and_key_id(ErlNifEnv *env, ERL_NIF_TERM key, char ** id, ENGINE **e) +{ + ERL_NIF_TERM engine_res, key_id_term; + struct engine_ctx *ctx; + ErlNifBinary key_id_bin; + + if (!enif_get_map_value(env, key, atom_engine, &engine_res) || + !enif_get_resource(env, engine_res, engine_ctx_rtype, (void**)&ctx) || + !enif_get_map_value(env, key, atom_key_id, &key_id_term) || + !enif_inspect_binary(env, key_id_term, &key_id_bin)) { + return 0; + } + else { + *e = ctx->engine; + return zero_terminate(key_id_bin, id); + } +} + + +static char *get_key_password(ErlNifEnv *env, ERL_NIF_TERM key) { + ERL_NIF_TERM tmp_term; + ErlNifBinary pwd_bin; + char *pwd; + if (enif_get_map_value(env, key, atom_password, &tmp_term) && + enif_inspect_binary(env, tmp_term, &pwd_bin) && + zero_terminate(pwd_bin, &pwd) + ) return pwd; + + return NULL; +} + +static int zero_terminate(ErlNifBinary bin, char **buf) { + *buf = enif_alloc(bin.size+1); + if (!*buf) + return 0; + memcpy(*buf, bin.data, bin.size); + *(*buf+bin.size) = 0; + return 1; +} +#endif + static int get_pkey_private_key(ErlNifEnv *env, ERL_NIF_TERM algorithm, ERL_NIF_TERM key, EVP_PKEY **pkey) { - if (algorithm == atom_rsa) { + if (enif_is_map(env, key)) { +#ifdef HAS_ENGINE_SUPPORT + /* Use key stored in engine */ + ENGINE *e; + char *id; + char *password; + + if (!get_engine_and_key_id(env, key, &id, &e)) + return PKEY_BADARG; + password = get_key_password(env, key); + *pkey = ENGINE_load_private_key(e, id, NULL, password); + if (!pkey) + return PKEY_BADARG; + enif_free(id); +#else + return PKEY_BADARG; +#endif + } + else if (algorithm == atom_rsa) { RSA *rsa = RSA_new(); if (!get_rsa_private_key(env, key, rsa)) { @@ -3866,7 +4079,24 @@ static int get_pkey_private_key(ErlNifEnv *env, ERL_NIF_TERM algorithm, ERL_NIF_ static int get_pkey_public_key(ErlNifEnv *env, ERL_NIF_TERM algorithm, ERL_NIF_TERM key, EVP_PKEY **pkey) { - if (algorithm == atom_rsa) { + if (enif_is_map(env, key)) { +#ifdef HAS_ENGINE_SUPPORT + /* Use key stored in engine */ + ENGINE *e; + char *id; + char *password; + + if (!get_engine_and_key_id(env, key, &id, &e)) + return PKEY_BADARG; + password = get_key_password(env, key); + *pkey = ENGINE_load_public_key(e, id, NULL, password); + if (!pkey) + return PKEY_BADARG; + enif_free(id); +#else + return PKEY_BADARG; +#endif + } else if (algorithm == atom_rsa) { RSA *rsa = RSA_new(); if (!get_rsa_public_key(env, key, rsa)) { @@ -3924,7 +4154,7 @@ static int get_pkey_public_key(ErlNifEnv *env, ERL_NIF_TERM algorithm, ERL_NIF_T } static ERL_NIF_TERM pkey_sign_nif(ErlNifEnv *env, int argc, const ERL_NIF_TERM argv[]) -{/* (Algorithm, Type, Data|{digest,Digest}, Key, Options) */ +{/* (Algorithm, Type, Data|{digest,Digest}, Key|#{}, Options) */ int i; const EVP_MD *md = NULL; unsigned char md_value[EVP_MAX_MD_SIZE]; @@ -3944,6 +4174,13 @@ enif_get_atom(env,argv[0],buf,1024,ERL_NIF_LATIN1); printf("algo=%s ",buf); enif_get_atom(env,argv[1],buf,1024,ERL_NIF_LATIN1); printf("hash=%s ",buf); printf("\r\n"); */ + +#ifndef HAS_ENGINE_SUPPORT + if (enif_is_map(env, argv[3])) { + return atom_notsup; + } +#endif + i = get_pkey_sign_digest(env, argv[0], argv[1], argv[2], md_value, &md, &tbs, &tbslen); if (i != PKEY_OK) { if (i == PKEY_NOTSUP) @@ -3965,10 +4202,9 @@ printf("\r\n"); } #ifdef HAS_EVP_PKEY_CTX -/* printf("EVP interface\r\n"); - */ ctx = EVP_PKEY_CTX_new(pkey, NULL); if (!ctx) goto badarg; + if (EVP_PKEY_sign_init(ctx) <= 0) goto badarg; if (md != NULL && EVP_PKEY_CTX_set_signature_md(ctx, md) <= 0) goto badarg; @@ -4070,6 +4306,12 @@ static ERL_NIF_TERM pkey_verify_nif(ErlNifEnv *env, int argc, const ERL_NIF_TERM unsigned char *tbs; /* data to be signed */ size_t tbslen; +#ifndef HAS_ENGINE_SUPPORT + if (enif_is_map(env, argv[4])) { + return atom_notsup; + } +#endif + if (!enif_inspect_binary(env, argv[3], &sig_bin)) { return enif_make_badarg(env); } @@ -4095,7 +4337,7 @@ static ERL_NIF_TERM pkey_verify_nif(ErlNifEnv *env, int argc, const ERL_NIF_TERM } #ifdef HAS_EVP_PKEY_CTX -/* printf("EVP interface\r\n"); +/* printf("EVP interface\r\n"); */ ctx = EVP_PKEY_CTX_new(pkey, NULL); if (!ctx) goto badarg; @@ -4280,7 +4522,13 @@ static ERL_NIF_TERM pkey_crypt_nif(ErlNifEnv *env, int argc, const ERL_NIF_TERM int algo_init = 0; /* char algo[1024]; */ - + +#ifndef HAS_ENGINE_SUPPORT + if (enif_is_map(env, argv[2])) { + return atom_notsup; + } +#endif + if (!enif_inspect_binary(env, argv[1], &in_bin)) { return enif_make_badarg(env); } @@ -4542,6 +4790,83 @@ static ERL_NIF_TERM pkey_crypt_nif(ErlNifEnv *env, int argc, const ERL_NIF_TERM /*--------------------------------*/ +static ERL_NIF_TERM privkey_to_pubkey_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{ /* (Algorithm, PrivKey | KeyMap) */ + EVP_PKEY *pkey; + ERL_NIF_TERM alg = argv[0]; + ERL_NIF_TERM result[8]; + + if (get_pkey_private_key(env, alg, argv[1], &pkey) != PKEY_OK) { + return enif_make_badarg(env); + } + + if (alg == atom_rsa) { + const BIGNUM *n = NULL, *e = NULL, *d = NULL; + RSA *rsa = EVP_PKEY_get1_RSA(pkey); + if (rsa) { + RSA_get0_key(rsa, &n, &e, &d); + result[0] = bin_from_bn(env, e); // Exponent E + result[1] = bin_from_bn(env, n); // Modulus N = p*q + EVP_PKEY_free(pkey); + return enif_make_list_from_array(env, result, 2); + } + + } else if (argv[0] == atom_dss) { + const BIGNUM *p = NULL, *q = NULL, *g = NULL, *pub_key = NULL; + DSA *dsa = EVP_PKEY_get1_DSA(pkey); + if (dsa) { + DSA_get0_pqg(dsa, &p, &q, &g); + DSA_get0_key(dsa, &pub_key, NULL); + result[0] = bin_from_bn(env, p); + result[1] = bin_from_bn(env, q); + result[2] = bin_from_bn(env, g); + result[3] = bin_from_bn(env, pub_key); + EVP_PKEY_free(pkey); + return enif_make_list_from_array(env, result, 4); + } + + } else if (argv[0] == atom_ecdsa) { +#if defined(HAVE_EC) + EC_KEY *ec = EVP_PKEY_get1_EC_KEY(pkey); + if (ec) { + /* Example of result: + { + Curve = {Field, Prime, Point, Order, CoFactor} = + { + Field = {prime_field,<<255,...,255>>}, + Prime = {<<255,...,252>>, + <<90,...,75>>, + <<196,...,144>> + }, + Point = <<4,...,245>>, + Order = <<255,...,81>>, + CoFactor = <<1>> + }, + Key = <<151,...,62>> + } + or + { + Curve = + {characteristic_two_field, + M, + Basis = {tpbasis, _} + | {ppbasis, k1, k2, k3} + }, + Key + } + */ + EVP_PKEY_free(pkey); + return atom_notsup; + } +#else + EVP_PKEY_free(pkey); + return atom_notsup; +#endif + } + + if (pkey) EVP_PKEY_free(pkey); + return enif_make_badarg(env); +} /*================================================================*/ @@ -4554,3 +4879,598 @@ static ERL_NIF_TERM rand_seed_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM a RAND_seed(seed_bin.data,seed_bin.size); return atom_ok; } + +/*================================================================*/ +/* Engine */ +/*================================================================*/ +static ERL_NIF_TERM engine_by_id_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (EngineId) */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM ret; + ErlNifBinary engine_id_bin; + unsigned int engine_id_len = 0; + char *engine_id; + ENGINE *engine; + struct engine_ctx *ctx; + + // Get Engine Id + if(!enif_inspect_binary(env, argv[0], &engine_id_bin)) { + PRINTF_ERR0("engine_by_id_nif Leaved: badarg"); + return enif_make_badarg(env); + } else { + engine_id_len = engine_id_bin.size+1; + engine_id = enif_alloc(engine_id_len); + (void) memcpy(engine_id, engine_id_bin.data, engine_id_len); + engine_id[engine_id_len-1] = '\0'; + } + + engine = ENGINE_by_id(engine_id); + if(!engine) { + PRINTF_ERR0("engine_by_id_nif Leaved: {error, bad_engine_id}"); + return enif_make_tuple2(env, atom_error, atom_bad_engine_id); + } + + ctx = enif_alloc_resource(engine_ctx_rtype, sizeof(struct engine_ctx)); + ctx->engine = engine; + ctx->id = engine_id; + + ret = enif_make_resource(env, ctx); + enif_release_resource(ctx); + + return enif_make_tuple2(env, atom_ok, ret); +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_init_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM ret = atom_ok; + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_init_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + if (!ENGINE_init(ctx->engine)) { + //ERR_print_errors_fp(stderr); + PRINTF_ERR0("engine_init_nif Leaved: {error, engine_init_failed}"); + return enif_make_tuple2(env, atom_error, atom_engine_init_failed); + } + + return ret; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_free_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_free_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + ENGINE_free(ctx->engine); + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_finish_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_finish_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + ENGINE_finish(ctx->engine); + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_load_dynamic_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* () */ +#ifdef HAS_ENGINE_SUPPORT + ENGINE_load_dynamic(); + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_ctrl_cmd_strings_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine, Commands) */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM ret = atom_ok; + unsigned int cmds_len = 0; + char **cmds = NULL; + struct engine_ctx *ctx; + int i; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_ctrl_cmd_strings_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + PRINTF_ERR1("Engine Id: %s\r\n", ENGINE_get_id(ctx->engine)); + + // Get Command List + if(!enif_get_list_length(env, argv[1], &cmds_len)) { + PRINTF_ERR0("engine_ctrl_cmd_strings_nif Leaved: Bad Command List"); + return enif_make_badarg(env); + } else { + cmds_len *= 2; // Key-Value list from erlang + cmds = enif_alloc((cmds_len+1)*sizeof(char*)); + if(get_engine_load_cmd_list(env, argv[1], cmds, 0)) { + PRINTF_ERR0("engine_ctrl_cmd_strings_nif Leaved: Couldn't read Command List"); + ret = enif_make_badarg(env); + goto error; + } + } + + for(i = 0; i < cmds_len; i+=2) { + PRINTF_ERR2("Cmd: %s:%s\r\n", + cmds[i] ? cmds[i] : "(NULL)", + cmds[i+1] ? cmds[i+1] : "(NULL)"); + if(!ENGINE_ctrl_cmd_string(ctx->engine, cmds[i], cmds[i+1], 0)) { + PRINTF_ERR2("Command failed: %s:%s\r\n", + cmds[i] ? cmds[i] : "(NULL)", + cmds[i+1] ? cmds[i+1] : "(NULL)"); + //ENGINE_free(ctx->engine); + ret = enif_make_tuple2(env, atom_error, atom_ctrl_cmd_failed); + PRINTF_ERR0("engine_ctrl_cmd_strings_nif Leaved: {error, ctrl_cmd_failed}"); + goto error; + } +} + + error: + for(i = 0; cmds != NULL && cmds[i] != NULL; i++) + enif_free(cmds[i]); + return ret; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_add_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_add_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + if (!ENGINE_add(ctx->engine)) { + PRINTF_ERR0("engine_add_nif Leaved: {error, add_engine_failed}"); + return enif_make_tuple2(env, atom_error, atom_add_engine_failed); + } + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_remove_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_remove_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + if (!ENGINE_remove(ctx->engine)) { + PRINTF_ERR0("engine_remove_nif Leaved: {error, remove_engine_failed}"); + return enif_make_tuple2(env, atom_error, atom_remove_engine_failed); + } + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_register_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine, EngineMethod) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + unsigned int method; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_register_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + // Get Method + if (!enif_get_uint(env, argv[1], &method)) { + PRINTF_ERR0("engine_register_nif Leaved: Parameter Method not an uint"); + return enif_make_badarg(env); + } + + switch(method) + { +#ifdef ENGINE_METHOD_RSA + case ENGINE_METHOD_RSA: + if (!ENGINE_register_RSA(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_DSA + case ENGINE_METHOD_DSA: + if (!ENGINE_register_DSA(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_DH + case ENGINE_METHOD_DH: + if (!ENGINE_register_DH(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_RAND + case ENGINE_METHOD_RAND: + if (!ENGINE_register_RAND(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_ECDH + case ENGINE_METHOD_ECDH: + if (!ENGINE_register_ECDH(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_ECDSA + case ENGINE_METHOD_ECDSA: + if (!ENGINE_register_ECDSA(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_STORE + case ENGINE_METHOD_STORE: + if (!ENGINE_register_STORE(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_CIPHERS + case ENGINE_METHOD_CIPHERS: + if (!ENGINE_register_ciphers(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_DIGESTS + case ENGINE_METHOD_DIGESTS: + if (!ENGINE_register_digests(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_PKEY_METHS + case ENGINE_METHOD_PKEY_METHS: + if (!ENGINE_register_pkey_meths(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_PKEY_ASN1_METHS + case ENGINE_METHOD_PKEY_ASN1_METHS: + if (!ENGINE_register_pkey_asn1_meths(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif +#ifdef ENGINE_METHOD_EC + case ENGINE_METHOD_EC: + if (!ENGINE_register_EC(ctx->engine)) + return enif_make_tuple2(env, atom_error, atom_register_engine_failed); + break; +#endif + default: + return enif_make_tuple2(env, atom_error, atom_engine_method_not_supported); + break; + } + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_unregister_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine, EngineMethod) */ +#ifdef HAS_ENGINE_SUPPORT + struct engine_ctx *ctx; + unsigned int method; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_unregister_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + // Get Method + if (!enif_get_uint(env, argv[1], &method)) { + PRINTF_ERR0("engine_unregister_nif Leaved: Parameter Method not an uint"); + return enif_make_badarg(env); + } + + switch(method) + { +#ifdef ENGINE_METHOD_RSA + case ENGINE_METHOD_RSA: + ENGINE_unregister_RSA(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_DSA + case ENGINE_METHOD_DSA: + ENGINE_unregister_DSA(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_DH + case ENGINE_METHOD_DH: + ENGINE_unregister_DH(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_RAND + case ENGINE_METHOD_RAND: + ENGINE_unregister_RAND(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_ECDH + case ENGINE_METHOD_ECDH: + ENGINE_unregister_ECDH(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_ECDSA + case ENGINE_METHOD_ECDSA: + ENGINE_unregister_ECDSA(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_STORE + case ENGINE_METHOD_STORE: + ENGINE_unregister_STORE(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_CIPHERS + case ENGINE_METHOD_CIPHERS: + ENGINE_unregister_ciphers(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_DIGESTS + case ENGINE_METHOD_DIGESTS: + ENGINE_unregister_digests(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_PKEY_METHS + case ENGINE_METHOD_PKEY_METHS: + ENGINE_unregister_pkey_meths(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_PKEY_ASN1_METHS + case ENGINE_METHOD_PKEY_ASN1_METHS: + ENGINE_unregister_pkey_asn1_meths(ctx->engine); + break; +#endif +#ifdef ENGINE_METHOD_EC + case ENGINE_METHOD_EC: + ENGINE_unregister_EC(ctx->engine); + break; +#endif + default: + break; + } + return atom_ok; +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_get_first_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM ret; + ENGINE *engine; + ErlNifBinary engine_bin; + struct engine_ctx *ctx; + + engine = ENGINE_get_first(); + if(!engine) { + enif_alloc_binary(0, &engine_bin); + engine_bin.size = 0; + return enif_make_tuple2(env, atom_ok, enif_make_binary(env, &engine_bin)); + } + + ctx = enif_alloc_resource(engine_ctx_rtype, sizeof(struct engine_ctx)); + ctx->engine = engine; + ctx->id = NULL; + + ret = enif_make_resource(env, ctx); + enif_release_resource(ctx); + + return enif_make_tuple2(env, atom_ok, ret); +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_get_next_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM ret; + ENGINE *engine; + ErlNifBinary engine_bin; + struct engine_ctx *ctx, *next_ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_get_next_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + engine = ENGINE_get_next(ctx->engine); + if (!engine) { + enif_alloc_binary(0, &engine_bin); + engine_bin.size = 0; + return enif_make_tuple2(env, atom_ok, enif_make_binary(env, &engine_bin)); + } + + next_ctx = enif_alloc_resource(engine_ctx_rtype, sizeof(struct engine_ctx)); + next_ctx->engine = engine; + next_ctx->id = NULL; + + ret = enif_make_resource(env, next_ctx); + enif_release_resource(next_ctx); + + return enif_make_tuple2(env, atom_ok, ret); +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_get_id_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* (Engine) */ +#ifdef HAS_ENGINE_SUPPORT + ErlNifBinary engine_id_bin; + const char *engine_id; + int size; + struct engine_ctx *ctx; + + // Get Engine + if (!enif_get_resource(env, argv[0], engine_ctx_rtype, (void**)&ctx)) { + PRINTF_ERR0("engine_get_id_nif Leaved: Parameter not an engine resource object"); + return enif_make_badarg(env); + } + + engine_id = ENGINE_get_id(ctx->engine); + if (!engine_id) { + enif_alloc_binary(0, &engine_id_bin); + engine_id_bin.size = 0; + return enif_make_tuple2(env, atom_ok, enif_make_binary(env, &engine_id_bin)); + } + + size = strlen(engine_id); + enif_alloc_binary(size, &engine_id_bin); + engine_id_bin.size = size; + memcpy(engine_id_bin.data, engine_id, size); + + return enif_make_tuple2(env, atom_ok, enif_make_binary(env, &engine_id_bin)); +#else + return atom_notsup; +#endif +} + +static int get_engine_load_cmd_list(ErlNifEnv* env, const ERL_NIF_TERM term, char **cmds, int i) +{ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM head, tail; + const ERL_NIF_TERM *tmp_tuple; + ErlNifBinary tmpbin; + int arity; + char* tmpstr; + int tmplen = 0; + + if(!enif_is_empty_list(env, term)) { + if(!enif_get_list_cell(env, term, &head, &tail)) { + cmds[i] = NULL; + return -1; + } else { + if(!enif_get_tuple(env, head, &arity, &tmp_tuple) || arity != 2) { + cmds[i] = NULL; + return -1; + } else { + if(!enif_inspect_binary(env, tmp_tuple[0], &tmpbin)) { + cmds[i] = NULL; + return -1; + } else { + tmplen = tmpbin.size+1; + tmpstr = enif_alloc(tmplen); + (void) memcpy(tmpstr, tmpbin.data, tmplen); + tmpstr[tmplen-1] = '\0'; + cmds[i++] = tmpstr; + } + if(!enif_inspect_binary(env, tmp_tuple[1], &tmpbin)) { + cmds[i] = NULL; + return -1; + } else { + if(tmpbin.size == 0) + cmds[i++] = NULL; + else { + tmplen = tmpbin.size+1; + tmpstr = enif_alloc(tmplen); + (void) memcpy(tmpstr, tmpbin.data, tmplen); + tmpstr[tmplen-1] = '\0'; + cmds[i++] = tmpstr; + } + } + return get_engine_load_cmd_list(env, tail, cmds, i); + } + } + } else { + cmds[i] = NULL; + return 0; + } +#else + return atom_notsup; +#endif +} + +static ERL_NIF_TERM engine_get_all_methods_nif(ErlNifEnv* env, int argc, const ERL_NIF_TERM argv[]) +{/* () */ +#ifdef HAS_ENGINE_SUPPORT + ERL_NIF_TERM method_array[12]; + int i = 0; + +#ifdef ENGINE_METHOD_RSA + method_array[i++] = atom_engine_method_rsa; +#endif +#ifdef ENGINE_METHOD_DSA + method_array[i++] = atom_engine_method_dsa; +#endif +#ifdef ENGINE_METHOD_DH + method_array[i++] = atom_engine_method_dh; +#endif +#ifdef ENGINE_METHOD_RAND + method_array[i++] = atom_engine_method_rand; +#endif +#ifdef ENGINE_METHOD_ECDH + method_array[i++] = atom_engine_method_ecdh; +#endif +#ifdef ENGINE_METHOD_ECDSA + method_array[i++] = atom_engine_method_ecdsa; +#endif +#ifdef ENGINE_METHOD_STORE + method_array[i++] = atom_engine_method_store; +#endif +#ifdef ENGINE_METHOD_CIPHERS + method_array[i++] = atom_engine_method_ciphers; +#endif +#ifdef ENGINE_METHOD_DIGESTS + method_array[i++] = atom_engine_method_digests; +#endif +#ifdef ENGINE_METHOD_PKEY_METHS + method_array[i++] = atom_engine_method_pkey_meths; +#endif +#ifdef ENGINE_METHOD_PKEY_ASN1_METHS + method_array[i++] = atom_engine_method_pkey_asn1_meths; +#endif +#ifdef ENGINE_METHOD_EC + method_array[i++] = atom_engine_method_ec; +#endif + + return enif_make_list_from_array(env, method_array, i); +#else + return atom_notsup; +#endif +} diff --git a/lib/crypto/c_src/otp_test_engine.c b/lib/crypto/c_src/otp_test_engine.c new file mode 100644 index 0000000000..a66bee2ddf --- /dev/null +++ b/lib/crypto/c_src/otp_test_engine.c @@ -0,0 +1,262 @@ +/* + * %CopyrightBegin% + * + * Copyright Ericsson AB 2017-2017. All Rights Reserved. + * + * Licensed under the Apache License, Version 2.0 (the "License"); + * you may not use this file except in compliance with the License. + * You may obtain a copy of the License at + * + * http://www.apache.org/licenses/LICENSE-2.0 + * + * Unless required by applicable law or agreed to in writing, software + * distributed under the License is distributed on an "AS IS" BASIS, + * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + * See the License for the specific language governing permissions and + * limitations under the License. + * + * %CopyrightEnd% + */ + +#ifdef _WIN32 +#define OPENSSL_OPT_WINDLL +#endif +#include <stdio.h> +#include <string.h> + +#include <openssl/engine.h> +#include <openssl/md5.h> +#include <openssl/rsa.h> +#include <openssl/pem.h> + +#define PACKED_OPENSSL_VERSION(MAJ, MIN, FIX, P) \ + ((((((((MAJ << 8) | MIN) << 8 ) | FIX) << 8) | (P-'a'+1)) << 4) | 0xf) + +#define PACKED_OPENSSL_VERSION_PLAIN(MAJ, MIN, FIX) \ + PACKED_OPENSSL_VERSION(MAJ,MIN,FIX,('a'-1)) + +#if OPENSSL_VERSION_NUMBER < PACKED_OPENSSL_VERSION_PLAIN(1,1,0) \ + || defined(LIBRESSL_VERSION_NUMBER) +#define OLD +#endif + +static const char *test_engine_id = "MD5"; +static const char *test_engine_name = "MD5 test engine"; + +/* The callback that does the job of fetching keys on demand by the Engine */ +EVP_PKEY* test_key_load(ENGINE *er, const char *id, UI_METHOD *ui_method, void *callback_data); + + +static int test_init(ENGINE *e) { + printf("OTP Test Engine Initializatzion!\r\n"); + + /* Load all digest and cipher algorithms. Needed for password protected private keys */ + OpenSSL_add_all_algorithms(); + + return 111; +} + +static void add_test_data(unsigned char *md, unsigned int len) +{ + unsigned int i; + + for (i=0; i<len; i++) { + md[i] = (unsigned char)(i & 0xff); + } +} + +/* MD5 part */ +#undef data +#ifdef OLD +#define data(ctx) ((MD5_CTX *)ctx->md_data) +#endif + +static int test_engine_md5_init(EVP_MD_CTX *ctx) { + fprintf(stderr, "MD5 initialized\r\n"); +#ifdef OLD + return MD5_Init(data(ctx)); +#else + return 1; +#endif +} + +static int test_engine_md5_update(EVP_MD_CTX *ctx,const void *data, size_t count) +{ + fprintf(stderr, "MD5 update\r\n"); +#ifdef OLD + return MD5_Update(data(ctx), data, (size_t)count); +#else + return 1; +#endif +} + +static int test_engine_md5_final(EVP_MD_CTX *ctx,unsigned char *md) { +#ifdef OLD + int ret; + + fprintf(stderr, "MD5 final size of EVP_MD: %lu\r\n", sizeof(EVP_MD)); + ret = MD5_Final(md, data(ctx)); + + if (ret > 0) { + add_test_data(md, MD5_DIGEST_LENGTH); + } + return ret; +#else + fprintf(stderr, "MD5 final\r\n"); + add_test_data(md, MD5_DIGEST_LENGTH); + return 1; +#endif +} + +#ifdef OLD +static EVP_MD test_engine_md5_method= { + NID_md5, /* The name ID for MD5 */ + NID_undef, /* IGNORED: MD5 with private key encryption NID */ + MD5_DIGEST_LENGTH, /* Size of MD5 result, in bytes */ + 0, /* Flags */ + test_engine_md5_init, /* digest init */ + test_engine_md5_update, /* digest update */ + test_engine_md5_final, /* digest final */ + NULL, /* digest copy */ + NULL, /* digest cleanup */ + EVP_PKEY_NULL_method, /* IGNORED: pkey methods */ + MD5_CBLOCK, /* Internal blocksize, see rfc1321/md5.h */ + sizeof(EVP_MD *) + sizeof(MD5_CTX), + NULL, /* IGNORED: control function */ +}; +#endif + +static int test_digest_ids[] = {NID_md5}; + +static int test_engine_digest_selector(ENGINE *e, const EVP_MD **digest, + const int **nids, int nid) { + int ok = 1; + if (!digest) { + *nids = test_digest_ids; + fprintf(stderr, "Digest is empty! Nid:%d\r\n", nid); + return 2; + } + fprintf(stderr, "Digest no %d requested\r\n",nid); + if (nid == NID_md5) { +#ifdef OLD + *digest = &test_engine_md5_method; +#else + EVP_MD *md = EVP_MD_meth_new(NID_md5, NID_undef); + if (!md || + !EVP_MD_meth_set_result_size(md, MD5_DIGEST_LENGTH) || + !EVP_MD_meth_set_flags(md, 0) || + !EVP_MD_meth_set_init(md, test_engine_md5_init) || + !EVP_MD_meth_set_update(md, test_engine_md5_update) || + !EVP_MD_meth_set_final(md, test_engine_md5_final) || + !EVP_MD_meth_set_copy(md, NULL) || + !EVP_MD_meth_set_cleanup(md, NULL) || + !EVP_MD_meth_set_input_blocksize(md, MD5_CBLOCK) || + !EVP_MD_meth_set_app_datasize(md, sizeof(EVP_MD *) + sizeof(MD5_CTX)) || + !EVP_MD_meth_set_ctrl(md, NULL)) + { + ok = 0; + *digest = NULL; + } else + { + *digest = md; + } +#endif + } + else { + ok = 0; + *digest = NULL; + } + + return ok; +} + + +static int bind_helper(ENGINE * e, const char *id) +{ + if (!ENGINE_set_id(e, test_engine_id) || + !ENGINE_set_name(e, test_engine_name) || + !ENGINE_set_init_function(e, test_init) || + !ENGINE_set_digests(e, &test_engine_digest_selector) || + /* For testing of key storage in an Engine: */ + !ENGINE_set_load_privkey_function(e, &test_key_load) || + !ENGINE_set_load_pubkey_function(e, &test_key_load) + ) + return 0; + + return 1; +} + +IMPLEMENT_DYNAMIC_CHECK_FN(); + +IMPLEMENT_DYNAMIC_BIND_FN(bind_helper); + +/******************************************************** + * + * Engine storage simulation + * + */ +int pem_passwd_cb_fun(char *buf, int size, int rwflag, void *password); + +EVP_PKEY* test_key_load(ENGINE *er, const char *id, UI_METHOD *ui_method, void *callback_data) +{ + EVP_PKEY *pkey = NULL; + FILE *f = fopen(id, "r"); + + if (!f) { + fprintf(stderr, "%s:%d fopen(%s) failed\r\n", __FILE__,__LINE__,id); + return NULL; + } + + /* First try to read as a private key. If that fails, try to read as a public key: */ + pkey = PEM_read_PrivateKey(f, NULL, pem_passwd_cb_fun, callback_data); + if (!pkey) { + /* ERR_print_errors_fp (stderr); */ + fclose(f); + f = fopen(id, "r"); + pkey = PEM_read_PUBKEY(f, NULL, NULL, NULL); + } + fclose(f); + + if (!pkey) { + fprintf(stderr, "%s:%d Key read from file failed. ", __FILE__,__LINE__); + if (callback_data) + fprintf(stderr, "Pwd = \"%s\". ", (char *)callback_data); + fprintf(stderr, "Contents of file \"%s\":\r\n",id); + f = fopen(id, "r"); + { /* Print the contents of the key file */ + char c; + while (!feof(f)) { + switch (c=fgetc(f)) { + case '\n': + case '\r': putc('\r',stdout); putc('\n',stdout); break; + default: putc(c, stdout); + } + } + } + fclose(f); + } + + return pkey; +} + + +int pem_passwd_cb_fun(char *buf, int size, int rwflag, void *password) +{ + int i; + + fprintf(stderr, "In pem_passwd_cb_fun\r\n"); + if (!password) + return 0; + + i = strlen(password); + if (i < size) { + /* whole pwd (incl terminating 0) fits */ + fprintf(stderr, "Got FULL pwd %d(%d) chars\r\n", i, size); + memcpy(buf, (char*)password, i+1); + return i+1; + } else { + fprintf(stderr, "Got TO LONG pwd %d(%d) chars\r\n", i, size); + /* meaningless with a truncated password */ + return 0; + } +} diff --git a/lib/crypto/doc/src/Makefile b/lib/crypto/doc/src/Makefile index 9c503b8fe0..a902779383 100644 --- a/lib/crypto/doc/src/Makefile +++ b/lib/crypto/doc/src/Makefile @@ -9,11 +9,11 @@ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. # See the License for the specific language governing permissions and # limitations under the License. -# +# # The Initial Developer of the Original Code is Ericsson Utvecklings AB. # Portions created by Ericsson are Copyright 1999, Ericsson Utvecklings # AB. All Rights Reserved.'' -# +# # $Id$ # include $(ERL_TOP)/make/target.mk @@ -38,13 +38,13 @@ XML_APPLICATION_FILES = ref_man.xml XML_REF3_FILES = crypto.xml XML_REF6_FILES = crypto_app.xml -XML_PART_FILES = release_notes.xml usersguide.xml -XML_CHAPTER_FILES = notes.xml licenses.xml fips.xml +XML_PART_FILES = usersguide.xml +XML_CHAPTER_FILES = notes.xml licenses.xml fips.xml engine_load.xml BOOK_FILES = book.xml XML_FILES = $(BOOK_FILES) $(XML_APPLICATION_FILES) $(XML_REF3_FILES) $(XML_REF6_FILES) \ - $(XML_PART_FILES) $(XML_CHAPTER_FILES) + $(XML_PART_FILES) $(XML_CHAPTER_FILES) GIF_FILES = @@ -63,9 +63,9 @@ HTML_REF_MAN_FILE = $(HTMLDIR)/index.html TOP_PDF_FILE = $(PDFDIR)/$(APPLICATION)-$(VSN).pdf # ---------------------------------------------------- -# FLAGS +# FLAGS # ---------------------------------------------------- -XML_FLAGS += +XML_FLAGS += # ---------------------------------------------------- # Targets @@ -73,7 +73,6 @@ XML_FLAGS += $(HTMLDIR)/%.gif: %.gif $(INSTALL_DATA) $< $@ - docs: pdf html man $(TOP_PDF_FILE): $(XML_FILES) @@ -86,7 +85,7 @@ man: $(MAN3_FILES) $(MAN6_FILES) gifs: $(GIF_FILES:%=$(HTMLDIR)/%) -debug opt valgrind: +debug opt valgrind: clean clean_docs clean_tex: rm -rf $(HTMLDIR)/* @@ -97,7 +96,7 @@ clean clean_docs clean_tex: # ---------------------------------------------------- # Release Target -# ---------------------------------------------------- +# ---------------------------------------------------- include $(ERL_TOP)/make/otp_release_targets.mk release_docs_spec: docs @@ -114,4 +113,3 @@ release_docs_spec: docs release_spec: - diff --git a/lib/crypto/doc/src/crypto.xml b/lib/crypto/doc/src/crypto.xml index 8c30d3d50c..3ab8f9c832 100644 --- a/lib/crypto/doc/src/crypto.xml +++ b/lib/crypto/doc/src/crypto.xml @@ -11,7 +11,7 @@ Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at - + http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software @@ -19,7 +19,6 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. - </legalnotice> <title>crypto</title> @@ -68,11 +67,11 @@ <section> <title>DATA TYPES </title> - - <code>key_value() = integer() | binary() </code> + + <code>key_value() = integer() | binary() </code> <p>Always <c>binary()</c> when used as return value</p> - <code>rsa_public() = [key_value()] = [E, N] </code> + <code>rsa_public() = [key_value()] = [E, N] </code> <p> Where E is the public exponent and N is public modulus. </p> <code>rsa_private() = [key_value()] = [E, N, D] | [E, N, D, P1, P2, E1, E2, C] </code> @@ -85,7 +84,7 @@ <code>dss_public() = [key_value()] = [P, Q, G, Y] </code> <p>Where P, Q and G are the dss parameters and Y is the public key.</p> - <code>dss_private() = [key_value()] = [P, Q, G, X] </code> + <code>dss_private() = [key_value()] = [P, Q, G, X] </code> <p>Where P, Q and G are the dss parameters and X is the private key.</p> <code>srp_public() = key_value() </code> @@ -109,15 +108,16 @@ <code>ecdh_private() = key_value() </code> - <code>ecdh_params() = ec_named_curve() | ec_explicit_curve()</code> + <code>ecdh_params() = ec_named_curve() | ec_explicit_curve()</code> <code>ec_explicit_curve() = - {ec_field(), Prime :: key_value(), Point :: key_value(), Order :: integer(), CoFactor :: none | integer()} </code> + {ec_field(), Prime :: key_value(), Point :: key_value(), Order :: integer(), + CoFactor :: none | integer()} </code> <code>ec_field() = {prime_field, Prime :: integer()} | {characteristic_two_field, M :: integer(), Basis :: ec_basis()}</code> - <code>ec_basis() = {tpbasis, K :: non_neg_integer()} | + <code>ec_basis() = {tpbasis, K :: non_neg_integer()} | {ppbasis, K1 :: non_neg_integer(), K2 :: non_neg_integer(), K3 :: non_neg_integer()} | onbasis</code> @@ -136,16 +136,33 @@ See also <seealso marker="#supports-0">crypto:supports/0</seealso> </p> + <code>engine_key_ref() = #{engine := engine_ref(), + key_id := key_id(), + password => password()}</code> + + <code>engine_key_ref() = term()</code> + <p>The result of a call to <seealso marker="#engine_load-3">engine_load/3</seealso>. + </p> + + <code>key_id() = string() | binary()</code> + <p>Identifies the key to be used. The format depends on the loaded engine. It is passed to + the <c>ENGINE_load_(private|public)_key</c> functions in libcrypto. + </p> + + <code>password() = string() | binary()</code> + <p>The key's password + </p> + <code>stream_cipher() = rc4 | aes_ctr </code> - <code>block_cipher() = aes_cbc | aes_cfb8 | aes_cfb128 | aes_ige256 | blowfish_cbc | + <code>block_cipher() = aes_cbc | aes_cfb8 | aes_cfb128 | aes_ige256 | blowfish_cbc | blowfish_cfb64 | des_cbc | des_cfb | des3_cbc | des3_cfb | des_ede3 | rc2_cbc </code> - <code>aead_cipher() = aes_gcm | chacha20_poly1305 </code> + <code>aead_cipher() = aes_gcm | chacha20_poly1305 </code> - <code>stream_key() = aes_key() | rc4_key() </code> + <code>stream_key() = aes_key() | rc4_key() </code> - <code>block_key() = aes_key() | blowfish_key() | des_key()| des3_key() </code> + <code>block_key() = aes_key() | blowfish_key() | des_key()| des3_key() </code> <code>aes_key() = iodata() </code> <p>Key length is 128, 192 or 256 bits</p> @@ -174,13 +191,17 @@ Note that both md4 and md5 are recommended only for compatibility with existing applications. </p> <code> cipher_algorithms() = aes_cbc | aes_cfb8 | aes_cfb128 | aes_ctr | aes_gcm | - aes_ige256 | blowfish_cbc | blowfish_cfb64 | chacha20_poly1305 | des_cbc | des_cfb | - des3_cbc | des3_cfb | des_ede3 | rc2_cbc | rc4 </code> - <code> mac_algorithms() = hmac | cmac</code> - <code> public_key_algorithms() = rsa |dss | ecdsa | dh | ecdh | ec_gf2m</code> + aes_ige256 | blowfish_cbc | blowfish_cfb64 | chacha20_poly1305 | des_cbc | + des_cfb | des3_cbc | des3_cfb | des_ede3 | rc2_cbc | rc4 </code> + <code> mac_algorithms() = hmac | cmac</code> + <code> public_key_algorithms() = rsa |dss | ecdsa | dh | ecdh | ec_gf2m</code> <p>Note that ec_gf2m is not strictly a public key algorithm, but a restriction on what curves are supported with ecdsa and ecdh. </p> + <code>engine_method_type() = engine_method_rsa | engine_method_dsa | engine_method_dh | + engine_method_rand | engine_method_ecdh | engine_method_ecdsa | + engine_method_ciphers | engine_method_digests | engine_method_store | + engine_method_pkey_meths | engine_method_pkey_asn1_meths</code> </section> @@ -261,13 +282,13 @@ is not supported by the underlying OpenSSL implementation.</p> </desc> </func> - + <func> <name>bytes_to_integer(Bin) -> Integer </name> <fsummary>Convert binary representation, of an integer, to an Erlang integer.</fsummary> <type> <v>Bin = binary() - as returned by crypto functions</v> - + <v>Integer = integer() </v> </type> <desc> @@ -439,7 +460,7 @@ </type> <desc> <p>Updates the HMAC represented by <c>Context</c> using the given <c>Data</c>. <c>Context</c> - must have been generated using an HMAC init function (such as + must have been generated using an HMAC init function (such as <seealso marker="#hmac_init-2">hmac_init</seealso>). <c>Data</c> can be any length. <c>NewContext</c> must be passed into the next call to <c>hmac_update</c> or to one of the functions <seealso marker="#hmac_final-1">hmac_final</seealso> and @@ -580,7 +601,7 @@ <type> <v>Type = rsa</v> <v>CipherText = binary()</v> - <v>PrivateKey = rsa_private()</v> + <v>PrivateKey = rsa_private() | engine_key_ref()</v> <v>Padding = rsa_pkcs1_padding | rsa_pkcs1_oaep_padding | rsa_no_padding</v> <v>PlainText = binary()</v> </type> @@ -594,7 +615,22 @@ </p> </desc> </func> - + + <func> + <name>privkey_to_pubkey(Type, EnginePrivateKeyRef) -> PublicKey</name> + <fsummary>Fetches a public key from an Engine stored private key.</fsummary> + <type> + <v>Type = rsa | dss</v> + <v>EnginePrivateKeyRef = engine_key_ref()</v> + <v>PublicKey = rsa_public() | dss_public()</v> + </type> + <desc> + <p>Fetches the corresponding public key from a private key stored in an Engine. + The key must be of the type indicated by the Type parameter. + </p> + </desc> + </func> + <func> <name>private_encrypt(Type, PlainText, PrivateKey, Padding) -> CipherText</name> <fsummary>Encrypts PlainText using the private Key.</fsummary> @@ -605,7 +641,7 @@ than <c>byte_size(N)-11</c> if <c>rsa_pkcs1_padding</c> is used, and <c>byte_size(N)</c> if <c>rsa_no_padding</c> is used, where N is public modulus of the RSA key.</d> - <v>PrivateKey = rsa_private()</v> + <v>PrivateKey = rsa_private() | engine_key_ref()</v> <v>Padding = rsa_pkcs1_padding | rsa_no_padding</v> <v>CipherText = binary()</v> </type> @@ -624,7 +660,7 @@ <type> <v>Type = rsa</v> <v>CipherText = binary()</v> - <v>PublicKey = rsa_public() </v> + <v>PublicKey = rsa_public() | engine_key_ref()</v> <v>Padding = rsa_pkcs1_padding | rsa_no_padding</v> <v>PlainText = binary()</v> </type> @@ -649,7 +685,7 @@ than <c>byte_size(N)-11</c> if <c>rsa_pkcs1_padding</c> is used, and <c>byte_size(N)</c> if <c>rsa_no_padding</c> is used, where N is public modulus of the RSA key.</d> - <v>PublicKey = rsa_public()</v> + <v>PublicKey = rsa_public() | engine_key_ref()</v> <v>Padding = rsa_pkcs1_padding | rsa_pkcs1_oaep_padding | rsa_no_padding</v> <v>CipherText = binary()</v> </type> @@ -702,7 +738,7 @@ signed or it is the hashed value of "cleartext" i.e. the digest (plaintext).</d> <v>DigestType = rsa_digest_type() | dss_digest_type() | ecdsa_digest_type()</v> - <v>Key = rsa_private() | dss_private() | [ecdh_private(),ecdh_params()]</v> + <v>Key = rsa_private() | dss_private() | [ecdh_private(),ecdh_params()] | engine_key_ref()</v> <v>Options = sign_options()</v> </type> <desc> @@ -1014,7 +1050,7 @@ _FloatValue = rand:uniform(). % [0.0; 1.0[</pre> or it is the hashed value of "cleartext" i.e. the digest (plaintext).</d> <v>DigestType = rsa_digest_type() | dss_digest_type() | ecdsa_digest_type()</v> <v>Signature = binary()</v> - <v>Key = rsa_public() | dss_public() | [ecdh_public(),ecdh_params()]</v> + <v>Key = rsa_public() | dss_public() | [ecdh_public(),ecdh_params()] | engine_key_ref()</v> <v>Options = sign_options()</v> </type> <desc> @@ -1026,6 +1062,124 @@ _FloatValue = rand:uniform(). % [0.0; 1.0[</pre> </desc> </func> + <!-- Engine functions --> + <func> + <name>engine_get_all_methods() -> Result</name> + <fsummary>Return list of all possible engine methods</fsummary> + <type> + <v>Result = [EngineMethod::atom()]</v> + </type> + <desc> + <p> + Returns a list of all possible engine methods. + </p> + <p> + May throw exception notsup in case there is + no engine support in the underlying OpenSSL implementation. + </p> + <p> + See also the chapter <seealso marker="crypto:engine_load#engine_load">Engine Load</seealso> + in the User's Guide. + </p> + </desc> + </func> + + <func> + <name>engine_load(EngineId, PreCmds, PostCmds) -> Result</name> + <fsummary>Dynamical load an encryption engine</fsummary> + <type> + <v>EngineId = unicode:chardata()</v> + <v>PreCmds, PostCmds = [{unicode:chardata(), unicode:chardata()}]</v> + <v>Result = {ok, Engine::term()} | {error, Reason::term()}</v> + </type> + <desc> + <p> + Loads the OpenSSL engine given by <c>EngineId</c> if it is available and then returns ok and + an engine handle. This function is the same as calling <c>engine_load/4</c> with + <c>EngineMethods</c> set to a list of all the possible methods. An error tuple is + returned if the engine can't be loaded. + </p> + <p> + The function throws a badarg if the parameters are in wrong format. + It may also throw the exception notsup in case there is + no engine support in the underlying OpenSSL implementation. + </p> + <p> + See also the chapter <seealso marker="crypto:engine_load#engine_load">Engine Load</seealso> + in the User's Guide. + </p> + </desc> + </func> + + <func> + <name>engine_load(EngineId, PreCmds, PostCmds, EngineMethods) -> Result</name> + <fsummary>Dynamical load an encryption engine</fsummary> + <type> + <v>EngineId = unicode:chardata()</v> + <v>PreCmds, PostCmds = [{unicode:chardata(), unicode:chardata()}]</v> + <v>EngineMethods = [engine_method_type()]</v> + <v>Result = {ok, Engine::term()} | {error, Reason::term()}</v> + </type> + <desc> + <p> + Loads the OpenSSL engine given by <c>EngineId</c> if it is available and then returns ok and + an engine handle. An error tuple is returned if the engine can't be loaded. + </p> + <p> + The function throws a badarg if the parameters are in wrong format. + It may also throw the exception notsup in case there is + no engine support in the underlying OpenSSL implementation. + </p> + <p> + See also the chapter <seealso marker="crypto:engine_load#engine_load">Engine Load</seealso> + in the User's Guide. + </p> + </desc> + </func> + + <func> + <name>engine_unload(Engine) -> Result</name> + <fsummary>Dynamical load an encryption engine</fsummary> + <type> + <v>Engine = term()</v> + <v>Result = ok | {error, Reason::term()}</v> + </type> + <desc> + <p> + Unloads the OpenSSL engine given by <c>EngineId</c>. + An error tuple is returned if the engine can't be unloaded. + </p> + <p> + The function throws a badarg if the parameter is in wrong format. + It may also throw the exception notsup in case there is + no engine support in the underlying OpenSSL implementation. + </p> + <p> + See also the chapter <seealso marker="crypto:engine_load#engine_load">Engine Load</seealso> + in the User's Guide. + </p> + </desc> + </func> + + <func> + <name>engine_list() -> Result</name> + <fsummary>List the known engine ids</fsummary> + <type> + <v>Result = [EngineId::unicode:chardata()]</v> + </type> + <desc> + <p>List the id's of all engines in OpenSSL's internal list.</p> + <p> + It may also throw the exception notsup in case there is + no engine support in the underlying OpenSSL implementation. + </p> + <p> + See also the chapter <seealso marker="crypto:engine_load#engine_load">Engine Load</seealso> + in the User's Guide. + </p> + </desc> + </func> + </funcs> <!-- Maybe put this in the users guide --> @@ -1100,4 +1254,3 @@ _FloatValue = rand:uniform(). % [0.0; 1.0[</pre> <!-- </p> --> <!-- </section> --> </erlref> - diff --git a/lib/crypto/doc/src/engine_load.xml b/lib/crypto/doc/src/engine_load.xml new file mode 100644 index 0000000000..e5c3f5d561 --- /dev/null +++ b/lib/crypto/doc/src/engine_load.xml @@ -0,0 +1,110 @@ +<?xml version="1.0" encoding="utf-8" ?> +<!DOCTYPE chapter SYSTEM "chapter.dtd"> + +<chapter> + <header> + <copyright> + <year>2017</year><year>2017</year> + <holder>Ericsson AB. All Rights Reserved.</holder> + </copyright> + <legalnotice> + The contents of this file are subject to the Erlang Public License, + Version 1.1, (the "License"); you may not use this file except in + compliance with the License. You should have received a copy of the + Erlang Public License along with this software. If not, it can be + retrieved online at http://www.erlang.org/. + + Software distributed under the License is distributed on an "AS IS" + basis, WITHOUT WARRANTY OF ANY KIND, either express or implied. See + the License for the specific language governing rights and limitations + under the License. + </legalnotice> + <title>Engine Load</title> + <prepared>Lars Thorsén</prepared> + <date>2017-08-22</date> + <file>engine_load.xml</file> + </header> + <p> + <marker id="engine_load"></marker> + This chapter describes the support for loading encryption engines in the crypto application. + </p> + + <section> + <title>Background</title> + <p> + OpenSSL exposes an Engine API, which makes it possible to plug in alternative + implementations for some or all of the cryptographic operations implemented by OpenSSL. + When configured appropriately, OpenSSL calls the engine's implementation of these + operations instead of its own. + </p> + <p> + Typically, OpenSSL engines provide a hardware implementation of specific cryptographic + operations. The hardware implementation usually offers improved performance over its + software-based counterpart, which is known as cryptographic acceleration. + </p> + </section> + + <section> + <title>Use Cases</title> + <section> + <title>Dynamically load an engine from default directory</title> + <p> + If the engine is located in the OpenSSL/LibreSSL installation <c>engines</c> directory. + </p> + <code> +1> {ok, Engine} = crypto:engine_load(<<"otp_test_engine">>, [], []). + {ok, #Ref}</code> + <note> + <p>The file name requirement on the engine dynamic library can differ between SSL versions.</p> + </note> + </section> + + <section> + <title>Load an engine with the dynamic engine</title> + <p> + Load an engine with the help of the dynamic engine by giving the path to the library. + </p> + <code> + 2> {ok, Engine} = crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, + <<"/some/path/otp_test_engine.so">>}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + []). + {ok, #Ref}</code> + <note> + <p>The dynamic engine is not supported in LibreSSL from version 2.2.1</p> + </note> + </section> + + <section> + <title>Load an engine and replace some methods</title> + <p> + Load an engine with the help of the dynamic engine and just + replace some engine methods. + </p> + <code> + 3> Methods = crypto:engine_get_all_methods() -- [engine_method_dh,engine_method_rand, +engine_method_ciphers,engine_method_digests, engine_method_store, +engine_method_pkey_meths, engine_method_pkey_asn1_meths]. +[engine_method_rsa,engine_method_dsa, + engine_method_ecdh,engine_method_ecdsa] + 4> {ok, Engine} = crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, + <<"/some/path/otp_test_engine.so">>}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + [], + Methods). + {ok, #Ref}</code> + </section> + + <section> + <title>List all engines currently loaded</title> + <code> + 5> crypto:engine_list(). +[<<"dynamic">>, <<"MD5">>]</code> + </section> + + </section> +</chapter> diff --git a/lib/crypto/doc/src/usersguide.xml b/lib/crypto/doc/src/usersguide.xml index 7971aefff4..f637a1db79 100644 --- a/lib/crypto/doc/src/usersguide.xml +++ b/lib/crypto/doc/src/usersguide.xml @@ -11,7 +11,7 @@ Licensed under the Apache License, Version 2.0 (the "License"); you may not use this file except in compliance with the License. You may obtain a copy of the License at - + http://www.apache.org/licenses/LICENSE-2.0 Unless required by applicable law or agreed to in writing, software @@ -19,7 +19,7 @@ WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. See the License for the specific language governing permissions and limitations under the License. - + </legalnotice> <title>Crypto User's Guide</title> @@ -48,5 +48,5 @@ </description> <xi:include href="licenses.xml"/> <xi:include href="fips.xml"/> + <xi:include href="engine_load.xml"/> </part> - diff --git a/lib/crypto/src/Makefile b/lib/crypto/src/Makefile index aea8a5a71c..edad0e6b61 100644 --- a/lib/crypto/src/Makefile +++ b/lib/crypto/src/Makefile @@ -39,8 +39,7 @@ MODULES= \ crypto \ crypto_ec_curves -HRL_FILES= - +HRL_FILES= ERL_FILES= $(MODULES:%=%.erl) TARGET_FILES= $(MODULES:%=$(EBIN)/%.$(EMULATOR)) @@ -56,16 +55,16 @@ APPUP_TARGET= $(EBIN)/$(APPUP_FILE) # ---------------------------------------------------- # FLAGS # ---------------------------------------------------- -ERL_COMPILE_FLAGS += -DCRYPTO_VSN=\"$(VSN)\" -Werror +ERL_COMPILE_FLAGS += -DCRYPTO_VSN=\"$(VSN)\" -Werror -I../include # ---------------------------------------------------- # Targets # ---------------------------------------------------- -debug opt valgrind: $(TARGET_FILES) $(APP_TARGET) $(APPUP_TARGET) +debug opt valgrind: $(TARGET_FILES) $(APP_TARGET) $(APPUP_TARGET) clean: - rm -f $(TARGET_FILES) $(APP_TARGET) $(APPUP_TARGET) + rm -f $(TARGET_FILES) $(APP_TARGET) $(APPUP_TARGET) rm -f errs core *~ $(APP_TARGET): $(APP_SRC) ../vsn.mk @@ -78,7 +77,7 @@ docs: # ---------------------------------------------------- # Release Target -# ---------------------------------------------------- +# ---------------------------------------------------- include $(ERL_TOP)/make/otp_release_targets.mk release_spec: opt @@ -89,10 +88,3 @@ release_spec: opt $(APPUP_TARGET) "$(RELSYSDIR)/ebin" release_docs_spec: - - - - - - - diff --git a/lib/crypto/src/crypto.erl b/lib/crypto/src/crypto.erl index 6b4f3a256d..9ba1623348 100644 --- a/lib/crypto/src/crypto.erl +++ b/lib/crypto/src/crypto.erl @@ -43,12 +43,26 @@ -export([public_encrypt/4, private_decrypt/4]). -export([private_encrypt/4, public_decrypt/4]). -export([dh_generate_parameters/2, dh_check/1]). %% Testing see +-export([privkey_to_pubkey/2]). -export([ec_curve/1, ec_curves/0]). -export([rand_seed/1]). +%% Engine +-export([ + engine_get_all_methods/0, + engine_load/3, + engine_load/4, + engine_unload/1, + engine_list/0 + ]). + +-export_type([engine_ref/0, + key_id/0, + password/0 + ]). -%% Private. For tests. --export([packed_openssl_version/4]). +%% Private. For tests. +-export([packed_openssl_version/4, engine_methods_convert_to_bitmask/2, get_test_engine/0]). -deprecated({rand_uniform, 2, next_major_release}). @@ -460,13 +474,24 @@ sign(Algorithm, Type, Data, Key, Options) -> end. + +-type key_id() :: string() | binary() . +-type password() :: string() | binary() . + +-type engine_key_ref() :: #{engine := engine_ref(), + key_id := key_id(), + password => password(), + term() => term() + }. + -type pk_algs() :: rsa | ecdsa | dss . --type pk_opt() :: list() | rsa_padding() . +-type pk_key() :: engine_key_ref() | [integer() | binary()] . +-type pk_opt() :: list() | rsa_padding() . --spec public_encrypt(pk_algs(), binary(), [binary()], pk_opt()) -> binary(). --spec public_decrypt(pk_algs(), binary(), [integer() | binary()], pk_opt()) -> binary(). --spec private_encrypt(pk_algs(), binary(), [integer() | binary()], pk_opt()) -> binary(). --spec private_decrypt(pk_algs(), binary(), [integer() | binary()], pk_opt()) -> binary(). +-spec public_encrypt(pk_algs(), binary(), pk_key(), pk_opt()) -> binary(). +-spec public_decrypt(pk_algs(), binary(), pk_key(), pk_opt()) -> binary(). +-spec private_encrypt(pk_algs(), binary(), pk_key(), pk_opt()) -> binary(). +-spec private_decrypt(pk_algs(), binary(), pk_key(), pk_opt()) -> binary(). public_encrypt(Algorithm, In, Key, Options) when is_list(Options) -> case pkey_crypt_nif(Algorithm, In, format_pkey(Algorithm, Key), Options, false, true) of @@ -607,10 +632,145 @@ compute_key(ecdh, Others, My, Curve) -> nif_curve_params(Curve), ensure_int_as_bin(My)). +%%====================================================================== +%% Engine functions +%%====================================================================== +%%---------------------------------------------------------------------- +%% Function: engine_get_all_methods/0 +%%---------------------------------------------------------------------- +-type engine_method_type() :: engine_method_rsa | engine_method_dsa | engine_method_dh | + engine_method_rand | engine_method_ecdh | engine_method_ecdsa | + engine_method_ciphers | engine_method_digests | engine_method_store | + engine_method_pkey_meths | engine_method_pkey_asn1_meths | + engine_method_ec. + +-type engine_ref() :: term(). + +-spec engine_get_all_methods() -> + [engine_method_type()]. +engine_get_all_methods() -> + notsup_to_error(engine_get_all_methods_nif()). + +%%---------------------------------------------------------------------- +%% Function: engine_load/3 +%%---------------------------------------------------------------------- +-spec engine_load(EngineId::unicode:chardata(), + PreCmds::[{unicode:chardata(), unicode:chardata()}], + PostCmds::[{unicode:chardata(), unicode:chardata()}]) -> + {ok, Engine::engine_ref()} | {error, Reason::term()}. +engine_load(EngineId, PreCmds, PostCmds) when is_list(PreCmds), is_list(PostCmds) -> + engine_load(EngineId, PreCmds, PostCmds, engine_get_all_methods()). + +%%---------------------------------------------------------------------- +%% Function: engine_load/4 +%%---------------------------------------------------------------------- +-spec engine_load(EngineId::unicode:chardata(), + PreCmds::[{unicode:chardata(), unicode:chardata()}], + PostCmds::[{unicode:chardata(), unicode:chardata()}], + EngineMethods::[engine_method_type()]) -> + {ok, Engine::term()} | {error, Reason::term()}. +engine_load(EngineId, PreCmds, PostCmds, EngineMethods) when is_list(PreCmds), + is_list(PostCmds) -> + try + ok = notsup_to_error(engine_load_dynamic_nif()), + case notsup_to_error(engine_by_id_nif(ensure_bin_chardata(EngineId))) of + {ok, Engine} -> + ok = engine_load_1(Engine, PreCmds, PostCmds, EngineMethods), + {ok, Engine}; + {error, Error1} -> + {error, Error1} + end + catch + throw:Error2 -> + Error2 + end. + +engine_load_1(Engine, PreCmds, PostCmds, EngineMethods) -> + try + ok = engine_nif_wrapper(engine_ctrl_cmd_strings_nif(Engine, ensure_bin_cmds(PreCmds))), + ok = engine_nif_wrapper(engine_add_nif(Engine)), + ok = engine_nif_wrapper(engine_init_nif(Engine)), + engine_load_2(Engine, PostCmds, EngineMethods), + ok + catch + throw:Error -> + %% The engine couldn't initialise, release the structural reference + ok = engine_free_nif(Engine), + throw(Error) + end. + +engine_load_2(Engine, PostCmds, EngineMethods) -> + try + ok = engine_nif_wrapper(engine_ctrl_cmd_strings_nif(Engine, ensure_bin_cmds(PostCmds))), + [ok = engine_nif_wrapper(engine_register_nif(Engine, engine_method_atom_to_int(Method))) || + Method <- EngineMethods], + ok + catch + throw:Error -> + %% The engine registration failed, release the functional reference + ok = engine_finish_nif(Engine), + throw(Error) + end. + +%%---------------------------------------------------------------------- +%% Function: engine_unload/1 +%%---------------------------------------------------------------------- +-spec engine_unload(Engine::term()) -> + ok | {error, Reason::term()}. +engine_unload(Engine) -> + engine_unload(Engine, engine_get_all_methods()). + +-spec engine_unload(Engine::term(), EngineMethods::[engine_method_type()]) -> + ok | {error, Reason::term()}. +engine_unload(Engine, EngineMethods) -> + try + [ok = engine_nif_wrapper(engine_unregister_nif(Engine, engine_method_atom_to_int(Method))) || + Method <- EngineMethods], + ok = engine_nif_wrapper(engine_remove_nif(Engine)), + %% Release the functional reference from engine_init_nif + ok = engine_nif_wrapper(engine_finish_nif(Engine)), + %% Release the structural reference from engine_by_id_nif + ok = engine_nif_wrapper(engine_free_nif(Engine)) + catch + throw:Error -> + Error + end. + +%%---------------------------------------------------------------------- +%% Function: engine_list/0 +%%---------------------------------------------------------------------- +-spec engine_list() -> + [EngineId::binary()]. +engine_list() -> + case notsup_to_error(engine_get_first_nif()) of + {ok, <<>>} -> + []; + {ok, Engine} -> + case notsup_to_error(engine_get_id_nif(Engine)) of + {ok, <<>>} -> + engine_list(Engine, []); + {ok, EngineId} -> + engine_list(Engine, [EngineId]) + end + end. + +engine_list(Engine0, IdList) -> + case notsup_to_error(engine_get_next_nif(Engine0)) of + {ok, <<>>} -> + lists:reverse(IdList); + {ok, Engine1} -> + case notsup_to_error(engine_get_id_nif(Engine1)) of + {ok, <<>>} -> + engine_list(Engine1, IdList); + {ok, EngineId} -> + engine_list(Engine1, [EngineId |IdList]) + end + end. + + %%-------------------------------------------------------------------- %%% On load %%-------------------------------------------------------------------- - on_load() -> LibBaseName = "crypto", PrivDir = code:priv_dir(crypto), @@ -670,12 +830,12 @@ path2bin(Path) when is_list(Path) -> end. %%-------------------------------------------------------------------- -%%% Internal functions +%%% Internal functions %%-------------------------------------------------------------------- max_bytes() -> ?MAX_BYTES_TO_NIF. -notsup_to_error(notsup) -> +notsup_to_error(notsup) -> erlang:error(notsup); notsup_to_error(Other) -> Other. @@ -799,7 +959,7 @@ do_stream_decrypt({rc4, State0}, Data) -> %% -%% AES - in counter mode (CTR) with state maintained for multi-call streaming +%% AES - in counter mode (CTR) with state maintained for multi-call streaming %% -type ctr_state() :: { iodata(), binary(), binary(), integer() } | binary(). @@ -808,11 +968,11 @@ do_stream_decrypt({rc4, State0}, Data) -> { ctr_state(), binary() }. -spec aes_ctr_stream_decrypt(ctr_state(), binary()) -> { ctr_state(), binary() }. - + aes_ctr_stream_init(_Key, _IVec) -> ?nif_stub. aes_ctr_stream_encrypt(_State, _Data) -> ?nif_stub. aes_ctr_stream_decrypt(_State, _Cipher) -> ?nif_stub. - + %% %% RC4 - symmetric stream cipher %% @@ -897,22 +1057,22 @@ pkey_verify_nif(_Algorithm, _Type, _Data, _Signature, _Key, _Options) -> ?nif_st rsa_generate_key_nif(_Bits, _Exp) -> ?nif_stub. %% DH Diffie-Hellman functions -%% +%% %% Generate (and check) Parameters is not documented because they are implemented %% for testing (and offline parameter generation) only. -%% From the openssl doc: +%% From the openssl doc: %% DH_generate_parameters() may run for several hours before finding a suitable prime. -%% Thus dh_generate_parameters may in this implementation block +%% Thus dh_generate_parameters may in this implementation block %% the emulator for several hours. %% -%% usage: dh_generate_parameters(1024, 2 or 5) -> +%% usage: dh_generate_parameters(1024, 2 or 5) -> %% [Prime=mpint(), SharedGenerator=mpint()] dh_generate_parameters(PrimeLen, Generator) -> case dh_generate_parameters_nif(PrimeLen, Generator) of error -> erlang:error(generation_failed, [PrimeLen,Generator]); Ret -> Ret - end. + end. dh_generate_parameters_nif(_PrimeLen, _Generator) -> ?nif_stub. @@ -938,6 +1098,16 @@ ec_curves() -> ec_curve(X) -> crypto_ec_curves:curve(X). + +privkey_to_pubkey(Alg, EngineMap) when Alg == rsa; Alg == dss; Alg == ecdsa -> + case privkey_to_pubkey_nif(Alg, format_pkey(Alg,EngineMap)) of + [_|_]=L -> map_ensure_bin_as_int(L); + X -> X + end. + +privkey_to_pubkey_nif(_Alg, _EngineMap) -> ?nif_stub. + + %% %% EC %% @@ -1005,6 +1175,19 @@ ensure_int_as_bin(Int) when is_integer(Int) -> ensure_int_as_bin(Bin) -> Bin. +map_ensure_bin_as_int(List) when is_list(List) -> + lists:map(fun ensure_bin_as_int/1, List). + +ensure_bin_as_int(Bin) when is_binary(Bin) -> + bin_to_int(Bin); +ensure_bin_as_int(E) -> + E. + +format_pkey(_Alg, #{engine:=_, key_id:=T}=M) when is_binary(T) -> format_pwd(M); +format_pkey(_Alg, #{engine:=_, key_id:=T}=M) when is_list(T) -> format_pwd(M#{key_id:=list_to_binary(T)}); +format_pkey(_Alg, #{engine:=_ }=M) -> error({bad_key_id, M}); +format_pkey(_Alg, #{}=M) -> error({bad_engine_map, M}); +%%% format_pkey(rsa, Key) -> map_ensure_int_as_bin(Key); format_pkey(ecdsa, [Key, Curve]) -> @@ -1014,6 +1197,9 @@ format_pkey(dss, Key) -> format_pkey(_, Key) -> Key. +format_pwd(#{password := Pwd}=M) when is_list(Pwd) -> M#{password := list_to_binary(Pwd)}; +format_pwd(M) -> M. + %%-------------------------------------------------------------------- %% -type rsa_padding() :: 'rsa_pkcs1_padding' | 'rsa_pkcs1_oaep_padding' | 'rsa_no_padding'. @@ -1024,7 +1210,7 @@ pkey_crypt_nif(_Algorithm, _In, _Key, _Options, _IsPrivate, _IsEncrypt) -> ?nif_ %% MP representaion (SSH2) mpint(X) when X < 0 -> mpint_neg(X); mpint(X) -> mpint_pos(X). - + -define(UINT32(X), X:32/unsigned-big-integer). @@ -1032,7 +1218,7 @@ mpint_neg(X) -> Bin = int_to_bin_neg(X, []), Sz = byte_size(Bin), <<?UINT32(Sz), Bin/binary>>. - + mpint_pos(X) -> Bin = int_to_bin_pos(X, []), <<MSB,_/binary>> = Bin, @@ -1054,7 +1240,6 @@ erlint(<<MPIntSize:32/integer,MPIntValue/binary>>) -> %% mod_exp_nif(_Base,_Exp,_Mod,_bin_hdr) -> ?nif_stub. - %%%---------------------------------------------------------------- %% 9470495 == V(0,9,8,zh). %% 268435615 == V(1,0,0,i). @@ -1065,3 +1250,92 @@ packed_openssl_version(MAJ, MIN, FIX, P0) -> P1 = atom_to_list(P0), P = lists:sum([C-$a||C<-P1]), ((((((((MAJ bsl 8) bor MIN) bsl 8 ) bor FIX) bsl 8) bor (P+1)) bsl 4) bor 16#f). + +%%-------------------------------------------------------------------- +%% Engine nifs +engine_by_id_nif(_EngineId) -> ?nif_stub. +engine_init_nif(_Engine) -> ?nif_stub. +engine_finish_nif(_Engine) -> ?nif_stub. +engine_free_nif(_Engine) -> ?nif_stub. +engine_load_dynamic_nif() -> ?nif_stub. +engine_ctrl_cmd_strings_nif(_Engine, _Cmds) -> ?nif_stub. +engine_add_nif(_Engine) -> ?nif_stub. +engine_remove_nif(_Engine) -> ?nif_stub. +engine_register_nif(_Engine, _EngineMethod) -> ?nif_stub. +engine_unregister_nif(_Engine, _EngineMethod) -> ?nif_stub. +engine_get_first_nif() -> ?nif_stub. +engine_get_next_nif(_Engine) -> ?nif_stub. +engine_get_id_nif(_Engine) -> ?nif_stub. +engine_get_all_methods_nif() -> ?nif_stub. + +%%-------------------------------------------------------------------- +%% Engine internals +engine_nif_wrapper(ok) -> + ok; +engine_nif_wrapper(notsup) -> + erlang:error(notsup); +engine_nif_wrapper({error, Error}) -> + throw({error, Error}). + +ensure_bin_chardata(CharData) when is_binary(CharData) -> + CharData; +ensure_bin_chardata(CharData) -> + unicode:characters_to_binary(CharData). + +ensure_bin_cmds(CMDs) -> + ensure_bin_cmds(CMDs, []). + +ensure_bin_cmds([], Acc) -> + lists:reverse(Acc); +ensure_bin_cmds([{Key, Value} |CMDs], Acc) -> + ensure_bin_cmds(CMDs, [{ensure_bin_chardata(Key), ensure_bin_chardata(Value)} | Acc]); +ensure_bin_cmds([Key | CMDs], Acc) -> + ensure_bin_cmds(CMDs, [{ensure_bin_chardata(Key), <<"">>} | Acc]). + +engine_methods_convert_to_bitmask([], BitMask) -> + BitMask; +engine_methods_convert_to_bitmask(engine_method_all, _BitMask) -> + 16#FFFF; +engine_methods_convert_to_bitmask(engine_method_none, _BitMask) -> + 16#0000; +engine_methods_convert_to_bitmask([M |Ms], BitMask) -> + engine_methods_convert_to_bitmask(Ms, BitMask bor engine_method_atom_to_int(M)). + +engine_method_atom_to_int(engine_method_rsa) -> 16#0001; +engine_method_atom_to_int(engine_method_dsa) -> 16#0002; +engine_method_atom_to_int(engine_method_dh) -> 16#0004; +engine_method_atom_to_int(engine_method_rand) -> 16#0008; +engine_method_atom_to_int(engine_method_ecdh) -> 16#0010; +engine_method_atom_to_int(engine_method_ecdsa) -> 16#0020; +engine_method_atom_to_int(engine_method_ciphers) -> 16#0040; +engine_method_atom_to_int(engine_method_digests) -> 16#0080; +engine_method_atom_to_int(engine_method_store) -> 16#0100; +engine_method_atom_to_int(engine_method_pkey_meths) -> 16#0200; +engine_method_atom_to_int(engine_method_pkey_asn1_meths) -> 16#0400; +engine_method_atom_to_int(engine_method_ec) -> 16#0800; +engine_method_atom_to_int(X) -> + erlang:error(badarg, [X]). + +get_test_engine() -> + Type = erlang:system_info(system_architecture), + LibDir = filename:join([code:priv_dir(crypto), "lib"]), + ArchDir = filename:join([LibDir, Type]), + case filelib:is_dir(ArchDir) of + true -> check_otp_test_engine(ArchDir); + false -> check_otp_test_engine(LibDir) + end. + +check_otp_test_engine(LibDir) -> + case filelib:wildcard("otp_test_engine*", LibDir) of + [] -> + {error, notexist}; + [LibName] -> + LibPath = filename:join(LibDir,LibName), + case filelib:is_file(LibPath) of + true -> + {ok, unicode:characters_to_binary(LibPath)}; + false -> + {error, notexist} + end + end. + diff --git a/lib/crypto/test/Makefile b/lib/crypto/test/Makefile index 138081d386..e046a25338 100644 --- a/lib/crypto/test/Makefile +++ b/lib/crypto/test/Makefile @@ -7,7 +7,8 @@ include $(ERL_TOP)/make/$(TARGET)/otp.mk MODULES = \ blowfish_SUITE \ - crypto_SUITE + crypto_SUITE \ + engine_SUITE ERL_FILES= $(MODULES:%=%.erl) @@ -27,7 +28,7 @@ RELSYSDIR = $(RELEASE_PATH)/crypto_test # FLAGS # ---------------------------------------------------- ERL_MAKE_FLAGS += -ERL_COMPILE_FLAGS += +ERL_COMPILE_FLAGS += +nowarn_export_all EBIN = . MAKE_EMAKE = $(wildcard $(ERL_TOP)/make/make_emakefile) @@ -77,7 +78,7 @@ release_spec: release_tests_spec: $(TEST_TARGET) $(INSTALL_DIR) "$(RELSYSDIR)" $(INSTALL_DATA) crypto.spec crypto.cover $(RELTEST_FILES) "$(RELSYSDIR)" - @tar cfh - crypto_SUITE_data | (cd "$(RELSYSDIR)"; tar xf -) + @tar cfh - *_SUITE_data | (cd "$(RELSYSDIR)"; tar xf -) chmod -R u+w "$(RELSYSDIR)" release_docs_spec: diff --git a/lib/crypto/test/crypto_SUITE.erl b/lib/crypto/test/crypto_SUITE.erl index 69f02d3da6..6dab459df6 100644 --- a/lib/crypto/test/crypto_SUITE.erl +++ b/lib/crypto/test/crypto_SUITE.erl @@ -198,7 +198,7 @@ init_per_suite(Config) -> %% This is NOT how you want to do seeding, it is just here %% to make the tests pass. Check your OS manual for how you %% really want to seed. - {H,M,L} = erlang:now(), + {H,M,L} = erlang:timestamp(), Bin = <<H:24,M:20,L:20>>, crypto:rand_seed(<< <<Bin/binary>> || _ <- lists:seq(1,16) >>), Config diff --git a/lib/crypto/test/engine_SUITE.erl b/lib/crypto/test/engine_SUITE.erl new file mode 100644 index 0000000000..72bd59f8ab --- /dev/null +++ b/lib/crypto/test/engine_SUITE.erl @@ -0,0 +1,513 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2017-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +%% + +-module(engine_SUITE). + +-include_lib("common_test/include/ct.hrl"). + +%% Note: This directive should only be used in test suites. +-compile(export_all). + +%%-------------------------------------------------------------------- +%% Common Test interface functions ----------------------------------- +%%-------------------------------------------------------------------- +suite() -> + [{ct_hooks,[ts_install_cth]}, + {timetrap,{seconds, 10}} + ]. + +all() -> + [ + get_all_possible_methods, + engine_load_all_methods, + engine_load_some_methods, + bad_arguments, + unknown_engine, + pre_command_fail_bad_value, + pre_command_fail_bad_key, + failed_engine_init, + {group, engine_stored_key} + ]. + +groups() -> + [{engine_stored_key, [], + [sign_verify_rsa, + sign_verify_dsa, + sign_verify_ecdsa, + sign_verify_rsa_pwd, + priv_encrypt_pub_decrypt_rsa, + priv_encrypt_pub_decrypt_rsa_pwd, + pub_encrypt_priv_decrypt_rsa, + pub_encrypt_priv_decrypt_rsa_pwd, + get_pub_from_priv_key_dsa, + get_pub_from_priv_key_ecdsa + ]}]. + + +init_per_suite(Config) -> + try crypto:start() of + ok -> + Config; + {error,{already_started,crypto}} -> + Config + catch _:_ -> + {skip, "Crypto did not start"} + end. +end_per_suite(_Config) -> + ok. + +%%-------------------------------------------------------------------- +init_per_group(engine_stored_key, Config) -> + case load_storage_engine(Config) of + {ok, E} -> + KeyDir = key_dir(Config), + [{storage_engine,E}, {storage_dir,KeyDir} | Config]; + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {error, notsup} -> + {skip, "Engine not supported on this OpenSSL version"}; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"}; + Other -> + ct:log("Engine load failed: ~p",[Other]), + {fail, "Engine load failed"} + end; +init_per_group(_Group, Config0) -> + Config0. + +end_per_group(engine_stored_key, Config) -> + case proplists:get_value(storage_engine, Config) of + undefined -> + ok; + E -> + ok = crypto:engine_unload(E) + end; +end_per_group(_, _) -> + ok. + +%%-------------------------------------------------------------------- +init_per_testcase(_Case, Config) -> + Config. +end_per_testcase(_Case, _Config) -> + ok. + +%%------------------------------------------------------------------------- +%% Test cases starts here. +%%------------------------------------------------------------------------- +get_all_possible_methods() -> + [{doc, "Just fetch all possible engine methods supported."}]. + +get_all_possible_methods(Config) when is_list(Config) -> + try + List = crypto:engine_get_all_methods(), + ct:log("crypto:engine_get_all_methods() -> ~p\n", [List]), + ok + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end. + +engine_load_all_methods()-> + [{doc, "Use a dummy md5 engine that does not implement md5" + "but rather returns a static binary to test that crypto:engine_load " + "functions works."}]. + +engine_load_all_methods(Config) when is_list(Config) -> + case crypto:get_test_engine() of + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {ok, Engine} -> + try + Md5Hash1 = <<106,30,3,246,166,222,229,158,244,217,241,179,50,232,107,109>>, + Md5Hash1 = crypto:hash(md5, "Don't panic"), + Md5Hash2 = <<0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15>>, + case crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + []) of + {ok, E} -> + case crypto:hash(md5, "Don't panic") of + Md5Hash1 -> + ct:fail(fail_to_load_still_original_engine); + Md5Hash2 -> + ok; + _ -> + ct:fail(fail_to_load_engine) + end, + ok = crypto:engine_unload(E), + case crypto:hash(md5, "Don't panic") of + Md5Hash2 -> + ct:fail(fail_to_unload_still_test_engine); + Md5Hash1 -> + ok; + _ -> + ct:fail(fail_to_unload_engine) + end; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"} + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end + end. + +engine_load_some_methods()-> + [{doc, "Use a dummy md5 engine that does not implement md5" + "but rather returns a static binary to test that crypto:engine_load " + "functions works."}]. + +engine_load_some_methods(Config) when is_list(Config) -> + case crypto:get_test_engine() of + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {ok, Engine} -> + try + Md5Hash1 = <<106,30,3,246,166,222,229,158,244,217,241,179,50,232,107,109>>, + Md5Hash1 = crypto:hash(md5, "Don't panic"), + Md5Hash2 = <<0,1,2,3,4,5,6,7,8,9,10,11,12,13,14,15>>, + EngineMethods = crypto:engine_get_all_methods() -- + [engine_method_dh,engine_method_rand, + engine_method_ciphers, engine_method_store, + engine_method_pkey_meths, engine_method_pkey_asn1_meths], + case crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + [], + EngineMethods) of + {ok, E} -> + case crypto:hash(md5, "Don't panic") of + Md5Hash1 -> + ct:fail(fail_to_load_engine_still_original); + Md5Hash2 -> + ok; + _ -> + ct:fail(fail_to_load_engine) + end, + ok = crypto:engine_unload(E), + case crypto:hash(md5, "Don't panic") of + Md5Hash2 -> + ct:fail(fail_to_unload_still_test_engine); + Md5Hash1 -> + ok; + _ -> + ct:fail(fail_to_unload_engine) + end; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"} + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end + end. + +%%------------------------------------------------------------------------- +%% Error cases +bad_arguments()-> + [{doc, "Test different arguments in bad format."}]. + +bad_arguments(Config) when is_list(Config) -> + case crypto:get_test_engine() of + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {ok, Engine} -> + try + try + crypto:engine_load(fail_engine, [], []) + catch + error:badarg -> + ok + end, + try + crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + 1, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + []) + catch + error:badarg -> + ok + end, + try + crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + {'ID', <<"MD5">>}, + <<"LOAD">>], + []) + catch + error:badarg -> + ok + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end + end. + +unknown_engine() -> + [{doc, "Try to load a non existent engine."}]. + +unknown_engine(Config) when is_list(Config) -> + try + {error, bad_engine_id} = crypto:engine_load(<<"fail_engine">>, [], []), + ok + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end. + +pre_command_fail_bad_value() -> + [{doc, "Test pre command due to bad value"}]. + +pre_command_fail_bad_value(Config) when is_list(Config) -> + DataDir = unicode:characters_to_binary(code:priv_dir(crypto)), + try + case crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, + <<DataDir/binary, <<"/libfail_engine.so">>/binary >>}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + []) of + {error, ctrl_cmd_failed} -> + ok; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"} + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end. + +pre_command_fail_bad_key() -> + [{doc, "Test pre command due to bad key"}]. + +pre_command_fail_bad_key(Config) when is_list(Config) -> + try + case crypto:get_test_engine() of + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {ok, Engine} -> + case crypto:engine_load(<<"dynamic">>, + [{<<"SO_WRONG_PATH">>, Engine}, + {<<"ID">>, <<"MD5">>}, + <<"LOAD">>], + []) of + {error, ctrl_cmd_failed} -> + ok; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"} + end + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end. + +failed_engine_init()-> + [{doc, "Test failing engine init due to missed pre command"}]. + +failed_engine_init(Config) when is_list(Config) -> + try + case crypto:get_test_engine() of + {error, notexist} -> + {skip, "OTP Test engine not found"}; + {ok, Engine} -> + case crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + {<<"ID">>, <<"MD5">>}], + []) of + {error, add_engine_failed} -> + ok; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"} + end + end + catch + error:notsup -> + {skip, "Engine not supported on this OpenSSL version"} + end. + +%%%---------------------------------------------------------------- +%%% Pub/priv key storage tests. Thoose are for testing the crypto.erl +%%% support for using priv/pub keys stored in an engine. + +sign_verify_rsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key.pem")}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key.pem")}, + sign_verify(rsa, sha, Priv, Pub). + +sign_verify_dsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "dsa_private_key.pem")}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "dsa_public_key.pem")}, + sign_verify(dss, sha, Priv, Pub). + +sign_verify_ecdsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "ecdsa_private_key.pem")}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "ecdsa_public_key.pem")}, + sign_verify(ecdsa, sha, Priv, Pub). + +sign_verify_rsa_pwd(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key_pwd.pem"), + password => "password"}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key_pwd.pem")}, + sign_verify(rsa, sha, Priv, Pub). + +priv_encrypt_pub_decrypt_rsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key.pem")}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key.pem")}, + priv_enc_pub_dec(rsa, Priv, Pub, rsa_pkcs1_padding). + +priv_encrypt_pub_decrypt_rsa_pwd(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key_pwd.pem"), + password => "password"}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key_pwd.pem")}, + priv_enc_pub_dec(rsa, Priv, Pub, rsa_pkcs1_padding). + +pub_encrypt_priv_decrypt_rsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key.pem")}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key.pem")}, + pub_enc_priv_dec(rsa, Pub, Priv, rsa_pkcs1_padding). + +pub_encrypt_priv_decrypt_rsa_pwd(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key.pem"), + password => "password"}, + Pub = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_public_key.pem")}, + pub_enc_priv_dec(rsa, Pub, Priv, rsa_pkcs1_padding). + +get_pub_from_priv_key_rsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "rsa_private_key.pem")}, + Pub = crypto:privkey_to_pubkey(rsa, Priv), + ct:log("rsa Pub = ~p",[Pub]), + sign_verify(rsa, sha, Priv, Pub). + +get_pub_from_priv_key_dsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "dsa_private_key.pem")}, + Pub = crypto:privkey_to_pubkey(dss, Priv), + ct:log("dsa Pub = ~p",[Pub]), + sign_verify(dss, sha, Priv, Pub). + +get_pub_from_priv_key_ecdsa(Config) -> + Priv = #{engine => engine_ref(Config), + key_id => key_id(Config, "ecdsa_private_key.pem")}, + Pub = crypto:privkey_to_pubkey(ecdsa, Priv), + case Pub of + notsup -> {skip, "ECDSA not implemented"}; + _ -> + ct:log("ecdsa Pub = ~p",[Pub]), + sign_verify(ecdsa, sha, Priv, Pub) + end. + +%%%================================================================ +%%% Help for engine_stored_pub_priv_keys* test cases +%%% +load_storage_engine(_Config) -> + case crypto:get_test_engine() of + {ok, Engine} -> + try crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + <<"LOAD">>], + []) + catch + error:notsup -> + {error, notsup} + end; + + {error, Error} -> + {error, Error} + end. + + +key_dir(Config) -> + DataDir = unicode:characters_to_binary(proplists:get_value(data_dir, Config)), + filename:join(DataDir, "pkcs8"). + + +engine_ref(Config) -> + proplists:get_value(storage_engine, Config). + +key_id(Config, File) -> + filename:join(proplists:get_value(storage_dir,Config), File). + +pubkey_alg_supported(Alg) -> + lists:member(Alg, + proplists:get_value(public_keys, crypto:supports())). + + +pub_enc_priv_dec(Alg, KeyEnc, KeyDec, Padding) -> + case pubkey_alg_supported(Alg) of + true -> + PlainText = <<"Hej på dig">>, + CryptoText = crypto:public_encrypt(Alg, PlainText, KeyEnc, Padding), + case crypto:private_decrypt(Alg, CryptoText, KeyDec, Padding) of + PlainText -> ok; + _ -> {fail, "Encrypt-decrypt error"} + end; + false -> + {skip, lists:concat([Alg," is not supported by cryptolib"])} + end. + +priv_enc_pub_dec(Alg, KeyEnc, KeyDec, Padding) -> + case pubkey_alg_supported(Alg) of + true -> + PlainText = <<"Hej på dig">>, + CryptoText = crypto:private_encrypt(Alg, PlainText, KeyEnc, Padding), + case crypto:public_decrypt(Alg, CryptoText, KeyDec, Padding) of + PlainText -> ok; + _ -> {fail, "Encrypt-decrypt error"} + end; + false -> + {skip, lists:concat([Alg," is not supported by cryptolib"])} + end. + +sign_verify(Alg, Sha, KeySign, KeyVerify) -> + case pubkey_alg_supported(Alg) of + true -> + PlainText = <<"Hej på dig">>, + Signature = crypto:sign(Alg, Sha, PlainText, KeySign), + case crypto:verify(Alg, Sha, PlainText, Signature, KeyVerify) of + true -> ok; + _ -> {fail, "Sign-verify error"} + end; + false -> + {skip, lists:concat([Alg," is not supported by cryptolib"])} + end. diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_private_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_private_key.pem new file mode 100644 index 0000000000..778ffac675 --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_private_key.pem @@ -0,0 +1,9 @@ +-----BEGIN PRIVATE KEY----- +MIIBSwIBADCCASwGByqGSM44BAEwggEfAoGBAMyitTMR7vPbpqyAXJpqnB0AhFwQ +F87IE+JKFl5bD/MSkhhRV5sM73HUU1ooXY0FjhZ+cdLUCATuZR5ta4ydANqWIcAB +gX3IwF1B4zf5SXEKTWkUYneL9dOKtiZLtoG28swrk8xMxwX+0fLHkltCEj6FiTW9 +PFrv8GmIfV6DjcI9AhUAqXWbb3RtoN9Ld28fVMhGZrj3LJUCgYEAwnxGHGBMpJaF +2w7zAw3jHjL8PMYlV6vnufGHQlwF0ZUXJxRsvagMb/X1qACTu2VPYEVoLQGM3cfH +EhHoQmvSXGAyTfR7Bmn3gf1n/s/DcFbdZduUCZ/rAyIrfd0eSbc1I+kZk85UCsKK +w/IYdlqcuYa4Cgm2TapT5uEMqH4jhzEEFgIULh8swEUWmU8aJNWsrWl4eCiuUUg= +-----END PRIVATE KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_public_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_public_key.pem new file mode 100644 index 0000000000..0fa5428828 --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/dsa_public_key.pem @@ -0,0 +1,12 @@ +-----BEGIN PUBLIC KEY----- +MIIBtzCCASwGByqGSM44BAEwggEfAoGBAMyitTMR7vPbpqyAXJpqnB0AhFwQF87I +E+JKFl5bD/MSkhhRV5sM73HUU1ooXY0FjhZ+cdLUCATuZR5ta4ydANqWIcABgX3I +wF1B4zf5SXEKTWkUYneL9dOKtiZLtoG28swrk8xMxwX+0fLHkltCEj6FiTW9PFrv +8GmIfV6DjcI9AhUAqXWbb3RtoN9Ld28fVMhGZrj3LJUCgYEAwnxGHGBMpJaF2w7z +Aw3jHjL8PMYlV6vnufGHQlwF0ZUXJxRsvagMb/X1qACTu2VPYEVoLQGM3cfHEhHo +QmvSXGAyTfR7Bmn3gf1n/s/DcFbdZduUCZ/rAyIrfd0eSbc1I+kZk85UCsKKw/IY +dlqcuYa4Cgm2TapT5uEMqH4jhzEDgYQAAoGAXPygOFYdeKgfLmuIC303cESYXvic +e2GNJomv8vaWLZmbLVVDfwA1fNsuF1hZkWw8f7aYaN9iZ3yl9u4Yk4TbJKkqfJqd +dgVt288SUqvi+NMHODUzYi9KAOXxupXffZSvdu54gKRaDuFTZ5XNcRqIJWGYlJYg +NVHF5FPZ4Bk2FYA= +-----END PUBLIC KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_private_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_private_key.pem new file mode 100644 index 0000000000..a45522064f --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_private_key.pem @@ -0,0 +1,8 @@ +-----BEGIN PRIVATE KEY----- +MIHuAgEAMBAGByqGSM49AgEGBSuBBAAjBIHWMIHTAgEBBEIBparGjr0KcdNrVM2J +G0mW5ltP1QyvxDqBMyWLWo3fruRZv6Qoohl5skd1u4O+KJoM/UrrSTOXI/MDR7NN +i1yl7O+hgYkDgYYABAG8K2XVsK0ahG9+HIIPwCO0pJY8ulwSTXwIjkCGyB2lpglh +8qJmRzuyGcfRTslv8wfv0sPlT9H9PKDvgrTUL7rvQQDdOODNgVPXSecUoXoPn+X+ +eqxs77bjx+A5x0t/i3m5PfkaNPh5MZ1H/bWuOOdj2ZXZw0R4rlVc0zVrgnPU8L8S +BQ== +-----END PRIVATE KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_public_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_public_key.pem new file mode 100644 index 0000000000..6d22fe43fe --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/ecdsa_public_key.pem @@ -0,0 +1,6 @@ +-----BEGIN PUBLIC KEY----- +MIGbMBAGByqGSM49AgEGBSuBBAAjA4GGAAQBvCtl1bCtGoRvfhyCD8AjtKSWPLpc +Ek18CI5AhsgdpaYJYfKiZkc7shnH0U7Jb/MH79LD5U/R/Tyg74K01C+670EA3Tjg +zYFT10nnFKF6D5/l/nqsbO+248fgOcdLf4t5uT35GjT4eTGdR/21rjjnY9mV2cNE +eK5VXNM1a4Jz1PC/EgU= +-----END PUBLIC KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key.pem new file mode 100644 index 0000000000..ea0e3d3958 --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key.pem @@ -0,0 +1,28 @@ +-----BEGIN PRIVATE KEY----- +MIIEvwIBADANBgkqhkiG9w0BAQEFAASCBKkwggSlAgEAAoIBAQCwwb0/ddXGXTFK +4FLxXdV6a/WJMSoPPS55RvZIAHFsiTtvPLbJ8LxDsZ6wSVZLN0/UQ4wdWn9jftyj +U5/IxBVG8XOtKimTMvm3/ZOzVLueGHBbrLYscRv9oL85ulTKHWgrZDu0lBX5JJTI +v5UTCErzJRQbka9DG1GaBgDb1PlXfkzBWMwfsBZmwoC77KvCcIGCgbW/XCY03TP2 +3Tg8drvpByMStddP2FQ4fZ91qFUzPu8uhZEsqSQTFlmhgGEx7dLlky0xvu62RuAD +RTpINpcWZtWDHTdssOqu653LwwqBY8lBopCZ/4Af8QR3ZYkQhen1YLEbVheXRuzI +LSCZIiJNAgMBAAECggEBAJH4/fxpqQkvr2Shy33Pu1xlyhnpw01gfn/jrcKasxEq +aC4eWup86E2TY3U8q4pkfIXU3uLi+O9HNpmflwargNLc1mY8uqb44ygiv5bLNEKE +9k2PXcdoBfC4jxPyoNFl5cBn/7LK1TazEjiTl15na9ZPWcLG1pG5/vMPYCgsQ1sP +8J3c4E3aaXIj9QceYxBprl490OCzieGyZlRipncz3g4UShRc/b4cycvDZOJpmAy4 +zbWTcBcSMPVPi5coF0K8UcimiqZkotfb/2RLc433i34IdsIXMM+brdq+g8rmjg5a ++oQPy02M6tFApBruEhAz8DGgaLtDY6MLtyZAt3SjXnUCgYEA1zLgamdTHOqrrmIi +eIQBnAJiyIfcY8B9SX1OsLGYFCHiPVwgUY35B2c7MavMsGcExJhtE+uxU7o5djtM +R6r9cRHOXJ6EQwa8OwzzPqbM17/YqNDeK39bc9WOFUqRWrhDhVMPy6z8rmZr73mG +IUC7mBNx/1GBdVYXIlsXzC96dI8CgYEA0kUAhz6I5nyPa70NDEUYHLHf3IW1BCmE +UoVbraSePJtIEY/IqFx7oDuFo30d4n5z+8ICCtyid1h/Cp3mf3akOiqltYUfgV1G +JgcEjKKYWEnO7cfFyO7LB7Y3GYYDJNy6EzVWPiwTGk9ZTfFJEESmHC45Unxgd17m +Dx/R58rFgWMCgYBQXQWFdtSI5fH7C1bIHrPjKNju/h2FeurOuObcAVZDnmu4cmD3 +U8d9xkVKxVeJQM99A1coq0nrdI3k4zwXP3mp8fZYjDHkPe2pN6rW6L9yiohEcsuk +/siON1/5/4DMmidM8LnjW9R45HLGWWGHpX7oyco2iJ+Jy/6Tq+T1MX3PbQKBgQCm +hdsbQJ0u3CrBSmFQ/E9SOlRt0r4+45pVuCOY6yweF2QF9HcXTtbhWQJHLclDHJ5C +Ha18aKuKFN3XzKFFBPKe1jOSBDGlQ/dQGnKx5fr8wMdObM3oiaTlIJuWbRmEUgJT +QARjDIi8Z2b0YUhZx+Q9oSXoe3PyVYehJrQX+/BavQKBgQCIr7Zp0rQPbfqcTL+M +OYHUoNcb14f9f8hXeXHQOqVpsGwxGdRQAU9wbx/4+obKB5xIkzBsVNcJwavisNja +hegnGjTB/9Hc4m+5bMGwH0bhS2eQO4o+YYM2ypDmFQqDLRfFUlZ5PVHffm/aA9+g +GanNBCsmtoHtV6CJ1UZ7NmBuIA== +-----END PRIVATE KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key_pwd.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key_pwd.pem new file mode 100644 index 0000000000..501662fc35 --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_private_key_pwd.pem @@ -0,0 +1,30 @@ +-----BEGIN ENCRYPTED PRIVATE KEY----- +MIIFDjBABgkqhkiG9w0BBQ0wMzAbBgkqhkiG9w0BBQwwDgQIh888Iq6gxuMCAggA +MBQGCCqGSIb3DQMHBAic/11YZ8Nt5gSCBMjG/Jb4qiMoBS50iQvHXqcETPE+0NBr +jhsn9w94LkdRBstMPAsoKmY98Er96Rnde/NfmqlU9CupKTkd7Ce5poBf72Y6KMED +cPURyjbGRFsu6x9skXB2obhyKYEqAEF2oQAg4Qbe5v1qXBIgDuC/NgiJnM+w2zCZ +LkHSZB2/NmcnvDzcgPF7TM8pTO23xCJ33m37qjfWvHsgocVqZmL9wQ4+wr/NMYjJ +pJvX1OHW1vBsZsXh40WchalYRSB1VeO368QfsE8coRJztqbMzdce9EQdMB6Q6jlO +cetd3moLIoMP4I7HW0/SgokbycTbRiYSvRyU1TGc2WbW6BrFZV24IckcnnVUFatf +6HKUcaYLG68dJcRgs5QMGkcmgVvlddENHFmHZlo0eym/xSiUl/AT8/5odscm6ML8 +wW5sneax+TF4J2eYmiN7yjAUCodXVTNYNDVKo6uUhntlymbM0o4UitVIbPIfTDHl +sxJAEZ7vpuPqeNMxUk6G6zipuEjqsVbnuFSBSZmgKiGYcifRPUmqqINa3DdS4WVx +xaPWdHbHVRD//ze3h/FsA+1lIE5q2kUE0xXseJA1ISog++kJp14XeaaL2j/tx3Ob +OsbcaOAD/IUw/ItDt9kn0qzfnar7sS0Wov8AmJQxHmH7Lm93jHTLM05yE0AR/eBr +Mig2ZdC+9OqVC+GPuBkRjSs8NpltQIDroz6EV9IMwPwXm0szSYoyoPLmlHJUdnLs +ZUef+au6hYkEJBrvuisagnq5eT/fCV3hsjD7yODebNU2CmBTo6X2PRx/xsBHRMWl +QkoM9PBdSCnKv6HpHl4pchuoqU2NpFjN0BCaad6aHfZSTnqgzK4bEh1oO6dI8/rB +/eh71JyFFG5J4xbpaqz5Su01V1iwU5leK5bDwqals4M4+ZGHGciou7qnXUmX2fJl +r6DlMUa/xy+A2ZG0NuZR05yk2oB3+KVNMgp6zFty3XaxwoNtc8GTLtLnBnIh2rlP +mE1+I65LRWwrNQalPeOAUrYuEzhyp2Df7a8Ykas5PUH7MGR/S0Ge/dLxtE2bJuK4 +znbLAsGhvo/SbNxYqIp6D4iDtd3va6yUGncy41paA/vTKFVvXZDrXcwJQYYCVOGT +OwdzNuozU8Dc7oxsd8oakfC46kvmVaOrGvZbm56PFfprcaL/Hslska5xxEni/eZe +WRxZbCBhAVqS1pn5zkDQVUe9uFlR/x39Qi01HIlKLBsjpSs6qQsFArMe8hgXmXLG +xP+dyVuOE18NzSewdEjeqSRKIM7Qi8EOjZsI4HdSRBY7bh9VhmaVXDZiCSf33TTE +3y8nimzQAeuGoYg6WqHmWWC2Qnpki2HlaIH/ayXEyQWkP/qvg61e8ovdg9Fy8JOO +0AacXVt5zj0q00AW5bKx7usi4NIjZedi86hUm6H19aBm7r86BKjwYTEI/GOcdrbV +9HC/8ayOimgwiAG3gq+aLioWym+Z6KnsbVd7XReVbvM/InQx54WA2y5im0A+/c67 +oQFFPV84XGX9waeqv/K4Wzkm6HW+qVAEM67482VGOf0PVrlQMno6dOotT/Y7ljoZ +2iz0LmN9yylJnLPDrr1i6gzbs5OhhUgbF5LI2YP2wWdCZTl/DrKSIvQZWl8U+tw3 +ciA= +-----END ENCRYPTED PRIVATE KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key.pem new file mode 100644 index 0000000000..d3fb5a2cc9 --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key.pem @@ -0,0 +1,9 @@ +-----BEGIN PUBLIC KEY----- +MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAsMG9P3XVxl0xSuBS8V3V +emv1iTEqDz0ueUb2SABxbIk7bzy2yfC8Q7GesElWSzdP1EOMHVp/Y37co1OfyMQV +RvFzrSopkzL5t/2Ts1S7nhhwW6y2LHEb/aC/ObpUyh1oK2Q7tJQV+SSUyL+VEwhK +8yUUG5GvQxtRmgYA29T5V35MwVjMH7AWZsKAu+yrwnCBgoG1v1wmNN0z9t04PHa7 +6QcjErXXT9hUOH2fdahVMz7vLoWRLKkkExZZoYBhMe3S5ZMtMb7utkbgA0U6SDaX +FmbVgx03bLDqruudy8MKgWPJQaKQmf+AH/EEd2WJEIXp9WCxG1YXl0bsyC0gmSIi +TQIDAQAB +-----END PUBLIC KEY----- diff --git a/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key_pwd.pem b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key_pwd.pem new file mode 100644 index 0000000000..f74361cead --- /dev/null +++ b/lib/crypto/test/engine_SUITE_data/pkcs8/rsa_public_key_pwd.pem @@ -0,0 +1,9 @@ +-----BEGIN PUBLIC KEY----- +MIIBIjANBgkqhkiG9w0BAQEFAAOCAQ8AMIIBCgKCAQEAxquo1Na8C+kjeW0YESGm +vE1bgNW9xh+SQjU1fv/97ePK8mQW2zO1h/vUNz23pfZAKjQu3rlFW/VgGJQ0LgCs +8Gr/HbMwNcCJzuFMePUrnWn/qBeR7OKUZCJ3E1pp4kwsTdGDDO7jPtNzKf0bdKlg +G2GHfZWhUediRX8NsRg12X1odVPuRGVRsyJ952YODk9PFjK7pro7Ynf3Icx7di9d +PXL5vEcKSRdomXvt1rgM8XVHES94RQqoz60ZhfV2JnPfa9V8qu0KaGntpEr7p4rQ +5BSiLFPjPOArjsD5tKyo8ldKCdQjLfisEp7AetfMjLPVVPw9o/SmCjDxsYWTVRQ2 +tQIDAQAB +-----END PUBLIC KEY----- diff --git a/lib/dialyzer/src/dialyzer_analysis_callgraph.erl b/lib/dialyzer/src/dialyzer_analysis_callgraph.erl index a4b42c9367..9993c68fed 100644 --- a/lib/dialyzer/src/dialyzer_analysis_callgraph.erl +++ b/lib/dialyzer/src/dialyzer_analysis_callgraph.erl @@ -165,7 +165,11 @@ analysis_start(Parent, Analysis, LegalWarnings) -> remote_type_postprocessing(TmpCServer, Args) -> Fun = fun() -> - exit(remote_type_postproc(TmpCServer, Args)) + exit(try remote_type_postproc(TmpCServer, Args) of + R -> R + catch + throw:{error,_}=Error -> Error + end) end, {Pid, Ref} = erlang:spawn_monitor(Fun), dialyzer_codeserver:give_away(TmpCServer, Pid), diff --git a/lib/dialyzer/test/plt_SUITE.erl b/lib/dialyzer/test/plt_SUITE.erl index ebe79b2a6d..a8a9f176fc 100644 --- a/lib/dialyzer/test/plt_SUITE.erl +++ b/lib/dialyzer/test/plt_SUITE.erl @@ -9,14 +9,14 @@ -export([suite/0, all/0, build_plt/1, beam_tests/1, update_plt/1, local_fun_same_as_callback/1, remove_plt/1, run_plt_check/1, run_succ_typings/1, - bad_dialyzer_attr/1, merge_plts/1]). + bad_dialyzer_attr/1, merge_plts/1, bad_record_type/1]). suite() -> [{timetrap, ?plt_timeout}]. all() -> [build_plt, beam_tests, update_plt, run_plt_check, remove_plt, run_succ_typings, local_fun_same_as_callback, - bad_dialyzer_attr, merge_plts]. + bad_dialyzer_attr, merge_plts, bad_record_type]. build_plt(Config) -> OutDir = ?config(priv_dir, Config), @@ -369,6 +369,32 @@ create_plts(Mod1, Mod2, Config) -> %% End of merge_plts(). +bad_record_type(Config) -> + PrivDir = ?config(priv_dir, Config), + Source = lists:concat([bad_record_type, ".erl"]), + Filename = filename:join(PrivDir, Source), + PltFilename = dialyzer_common:plt_file(PrivDir), + + Opts = [{files, [Filename]}, + {check_plt, false}, + {from, src_code}, + {init_plt, PltFilename}], + + Prog = <<"-module(bad_record_type). + -export([r/0]). + -record(r, {f = 3 :: integer()}). + -spec r() -> #r{f :: atom()}. + r() -> + #r{}.">>, + ok = file:write_file(Filename, Prog), + {dialyzer_error, + "Analysis failed with error:\n" ++ Str} = + (catch dialyzer:run(Opts)), + P = string:str(Str, + "bad_record_type.erl:4: Illegal declaration of #r{f}"), + true = P > 0, + ok. + erlang_beam() -> case code:where_is_file("erlang.beam") of non_existing -> diff --git a/lib/erl_docgen/priv/dtd/erlref.dtd b/lib/erl_docgen/priv/dtd/erlref.dtd index 615b88b61a..78d6771f52 100644 --- a/lib/erl_docgen/priv/dtd/erlref.dtd +++ b/lib/erl_docgen/priv/dtd/erlref.dtd @@ -33,4 +33,5 @@ <!ATTLIST name name CDATA #IMPLIED arity CDATA #IMPLIED clause_i CDATA #IMPLIED + anchor CDATA #IMPLIED n_vars CDATA #IMPLIED> diff --git a/lib/erl_docgen/priv/xsl/db_html.xsl b/lib/erl_docgen/priv/xsl/db_html.xsl index 75614392fb..5b7eae4f73 100644 --- a/lib/erl_docgen/priv/xsl/db_html.xsl +++ b/lib/erl_docgen/priv/xsl/db_html.xsl @@ -172,6 +172,7 @@ <xsl:template name="spec_name"> <xsl:variable name="name" select="@name"/> <xsl:variable name="arity" select="@arity"/> + <xsl:variable name="anchor" select="@anchor"/> <xsl:variable name="spec0"> <xsl:call-template name="find_spec"/> </xsl:variable> @@ -198,6 +199,11 @@ </xsl:otherwise> </xsl:choose> + <!-- Insert an anchor for "anchor" attribute --> + <xsl:if test="string-length($anchor) > 0"> + <a name="{$anchor}"></a> + </xsl:if> + <xsl:variable name="global_types" select="ancestor::erlref/datatypes"/> <xsl:variable name="local_types" select="../type[string-length(@name) > 0]"/> diff --git a/lib/hipe/doc/src/hipe_app.xml b/lib/hipe/doc/src/hipe_app.xml index e489d155c3..9299c6d73f 100644 --- a/lib/hipe/doc/src/hipe_app.xml +++ b/lib/hipe/doc/src/hipe_app.xml @@ -47,6 +47,72 @@ Details on HiPE compiler options are given by <c>hipe:help_options()</c>.</p> </description> <section> + <title>Feature Limitations</title> + <p> + The HiPE compiler is in general compliant with the normal BEAM compiler, + with respect to semantic behavior. There are however features in the BEAM compiler + and the runtime system that have limited or no support for HiPE compiled modules. + </p> + <taglist> + <tag>Stack traces</tag> + <item><p>Stack traces returned from <seealso marker="erts:erlang#get_stacktrace/0"> + <c>erlang:get_stacktrace/0</c></seealso> or as part of <c>'EXIT'</c> terms + can look incomplete if HiPE compiled functions are involved. Typically a stack trace + will contain only BEAM compiled functions or only HiPE compiled functions, depending + on where the exception was raised.</p> + <p>Source code line numbers in stack traces are also not supported by HiPE compiled functions.</p> + </item> + + <tag>Tracing</tag> + <item><p>Erlang call trace is not supported by HiPE. Calling + <seealso marker="erts:erlang#trace_pattern/3"><c>erlang:trace_pattern({M,F,A}, ...)</c></seealso> + does not have any effect on HiPE compiled modules.</p> + </item> + + <tag>NIFs</tag> + <item><p>Modules compiled with HiPE can not call <seealso marker="erts:erlang#load_nif-2"> + <c>erlang:load_nif/2</c></seealso> to load NIFs.</p> + </item> + + <tag>-on_load</tag> + <item><p>Modules compiled with HiPE can not use + <seealso marker="doc/reference_manual:code_loading#on_load"><c>-on_load()</c></seealso> + directives.</p> + </item> + </taglist> + + </section> + <section> + <title>Performance Limitations</title> + <p> + The HiPE compiler does in general produce faster code than the + BEAM compiler. There are however some situation when HiPE + compiled code will perform worse than BEAM code. + </p> + <taglist> + <tag>Mode switches</tag> + <item><p>Every time a process changes from executing code in a + HiPE compiled module to a BEAM compiled module (or vice versa), + it will do a mode switch. This involves a certain amount of + CPU overhead which can have a negative net impact if the + process is switching back and forth without getting enough done in + each mode.</p> + </item> + + <tag>Optimization for <c>receive</c> with unique references</tag> + <item><p>The BEAM compiler can do an optimization when + a <c>receive</c> statement is <em>only</em> waiting for messages + containing a reference created before the receive. All messages + that existed in the queue when the reference was created will be + bypassed, as they cannot possibly contain the reference. HiPE + does not implement this optimization.</p> + <p>An example of this is when + <c>gen_server:call()</c> waits for the reply message.</p> + </item> + + </taglist> + </section> + <section> <title>SEE ALSO</title> <p> <seealso marker="stdlib:c">c(3)</seealso>, diff --git a/lib/inets/doc/src/notes.xml b/lib/inets/doc/src/notes.xml index 1ff6aefaa7..07e29b5542 100644 --- a/lib/inets/doc/src/notes.xml +++ b/lib/inets/doc/src/notes.xml @@ -33,7 +33,26 @@ <file>notes.xml</file> </header> - <section><title>Inets 6.4.3</title> + <section><title>Inets 6.4.4</title> + + <section><title>Fixed Bugs and Malfunctions</title> + <list> + <item> + <p> + Correct the handling of location headers so that the + status code is not hard coded. This should have been + fixed by commit 2cc5ba70cbbc6b3ace81a2a0324417c3b65265bb + but unfortunately was broken during a code refactoring + and unnoticed due to a faulty placed test case.</p> + <p> + Own Id: OTP-14761</p> + </item> + </list> + </section> + +</section> + +<section><title>Inets 6.4.3</title> <section><title>Improvements and New Features</title> <list> diff --git a/lib/inets/src/http_server/httpd_esi.erl b/lib/inets/src/http_server/httpd_esi.erl index fd50934d00..f5493f6fad 100644 --- a/lib/inets/src/http_server/httpd_esi.erl +++ b/lib/inets/src/http_server/httpd_esi.erl @@ -86,7 +86,7 @@ handle_headers([], NewHeaders, StatusCode, _) -> handle_headers([Header | Headers], NewHeaders, StatusCode, NoESIStatus) -> {FieldName, FieldValue} = httpd_response:split_header(Header, []), case FieldName of - "location" -> + "location" when NoESIStatus == true -> handle_headers(Headers, [{FieldName, FieldValue} | NewHeaders], 302, NoESIStatus); diff --git a/lib/inets/src/http_server/httpd_example.erl b/lib/inets/src/http_server/httpd_example.erl index adbbf64685..47a8c48d01 100644 --- a/lib/inets/src/http_server/httpd_example.erl +++ b/lib/inets/src/http_server/httpd_example.erl @@ -91,7 +91,7 @@ yahoo(_Env,_Input) -> "Location: http://www.yahoo.com\r\n\r\n". new_status_and_location(_Env,_Input) -> - "status:201\r\n Location: http://www.yahoo.com\r\n\r\n". + "status:201 Created\r\n Location: http://www.yahoo.com\r\n\r\n". default(Env,Input) -> [header(), diff --git a/lib/inets/src/inets_app/inets.appup.src b/lib/inets/src/inets_app/inets.appup.src index a86413147c..fdf4cc6e07 100644 --- a/lib/inets/src/inets_app/inets.appup.src +++ b/lib/inets/src/inets_app/inets.appup.src @@ -18,10 +18,14 @@ %% %CopyrightEnd% {"%VSN%", [ + {<<"6.4.3">>, [{load_module, httpd_esi, + soft_purge, soft_purge, []}]}, {<<"6\\..*">>,[{restart_application, inets}]}, {<<"5\\..*">>,[{restart_application, inets}]} ], [ + {<<"6.4.3">>, [{load_module, httpd_esi, + soft_purge, soft_purge, []}]}, {<<"6\\..*">>,[{restart_application, inets}]}, {<<"5\\..*">>,[{restart_application, inets}]} ] diff --git a/lib/inets/test/http_format_SUITE.erl b/lib/inets/test/http_format_SUITE.erl index 9a13ed3d17..647eff4f7c 100644 --- a/lib/inets/test/http_format_SUITE.erl +++ b/lib/inets/test/http_format_SUITE.erl @@ -536,7 +536,10 @@ esi_parse_headers(Config) when is_list(Config) -> httpd_esi:handle_headers(Headers2), {ok,[{"location","/foo/bar.html"}], 302} = - httpd_esi:handle_headers("location:/foo/bar.html\r\n"). + httpd_esi:handle_headers("location:/foo/bar.html\r\n"), + + {ok,[{"location","http://foo/bar.html"}],201} = + httpd_esi:handle_headers("status:201 Created\r\nlocation:http://foo/bar.html\r\n"). %%-------------------------------------------------------------------- cgi_parse_headers() -> diff --git a/lib/inets/test/httpd_SUITE.erl b/lib/inets/test/httpd_SUITE.erl index 0c649d9abf..9a85c51d24 100644 --- a/lib/inets/test/httpd_SUITE.erl +++ b/lib/inets/test/httpd_SUITE.erl @@ -923,8 +923,11 @@ esi(Config) when is_list(Config) -> {no_header, "cache-control"}]), ok = http_status("GET /cgi-bin/erl/httpd_example:peer ", Config, [{statuscode, 200}, - {header, "peer-cert-exist", peer(Config)}]). - + {header, "peer-cert-exist", peer(Config)}]), + ok = http_status("GET /cgi-bin/erl/httpd_example:new_status_and_location ", + Config, [{statuscode, 201}, + {header, "location"}]). + %%------------------------------------------------------------------------- esi_put() -> [{doc, "Test mod_esi PUT"}]. diff --git a/lib/inets/vsn.mk b/lib/inets/vsn.mk index 108d259823..560d524bac 100644 --- a/lib/inets/vsn.mk +++ b/lib/inets/vsn.mk @@ -19,6 +19,6 @@ # %CopyrightEnd% APPLICATION = inets -INETS_VSN = 6.4.3 +INETS_VSN = 6.4.4 PRE_VSN = APP_VSN = "$(APPLICATION)-$(INETS_VSN)$(PRE_VSN)" diff --git a/lib/kernel/test/zlib_SUITE.erl b/lib/kernel/test/zlib_SUITE.erl index 7be7e503df..f203ef878f 100644 --- a/lib/kernel/test/zlib_SUITE.erl +++ b/lib/kernel/test/zlib_SUITE.erl @@ -166,7 +166,7 @@ api_deflateInit(Config) when is_list(Config) -> ?m(ok, zlib:deflateInit(Z12,default,deflated,-Wbits,8,default)), ?m(ok,zlib:close(Z11)), ?m(ok,zlib:close(Z12)) - end, lists:seq(8, 15)), + end, lists:seq(9, 15)), lists:foreach(fun(MemLevel) -> Z = zlib:open(), @@ -213,12 +213,46 @@ api_deflateReset(Config) when is_list(Config) -> %% Test deflateParams. api_deflateParams(Config) when is_list(Config) -> + Levels = [none, default, best_speed, best_compression] ++ lists:seq(0, 9), + Strategies = [filtered, huffman_only, rle, default], + Z1 = zlib:open(), ?m(ok, zlib:deflateInit(Z1, default)), - ?m(L when is_list(L), zlib:deflate(Z1, <<1,1,1,1,1,1,1,1,1>>, none)), - ?m(ok, zlib:deflateParams(Z1, best_compression, huffman_only)), - ?m(L when is_list(L), zlib:deflate(Z1, <<1,1,1,1,1,1,1,1,1>>, sync)), - ?m(ok, zlib:close(Z1)). + + ApiTest = + fun(Level, Strategy) -> + ?m(ok, zlib:deflateParams(Z1, Level, Strategy)), + ?m(ok, zlib:deflateReset(Z1)) + end, + + [ ApiTest(Level, Strategy) || Level <- Levels, Strategy <- Strategies ], + + ?m(ok, zlib:close(Z1)), + + FlushTest = + fun FlushTest(Size, Level, Strategy) -> + Z = zlib:open(), + ok = zlib:deflateInit(Z, default), + Data = gen_determ_rand_bytes(Size), + case zlib:deflate(Z, Data, none) of + [<<120, 156>>] -> + %% All data is present in the internal zlib state, and will + %% be flushed on deflateParams. + + ok = zlib:deflateParams(Z, Level, Strategy), + Compressed = [<<120, 156>>, zlib:deflate(Z, <<>>, finish)], + Data = zlib:uncompress(Compressed), + zlib:close(Z), + + FlushTest(Size + (1 bsl 10), Level, Strategy); + _Other -> + ok + end + end, + + [ FlushTest(1, Level, Strategy) || Level <- Levels, Strategy <- Strategies ], + + ok. %% Test deflate. api_deflate(Config) when is_list(Config) -> @@ -652,6 +686,11 @@ api_g_un_zip(Config) when is_list(Config) -> Concatenated = <<Bin/binary, Bin/binary>>, ?m(Concatenated, zlib:gunzip([Comp, Comp])), + %% Don't explode if the uncompressed size is a perfect multiple of the + %% internal inflate chunk size. + ChunkSizedData = <<0:16384/unit:8>>, + ?m(ChunkSizedData, zlib:gunzip(zlib:gzip(ChunkSizedData))), + %% Bad CRC; bad length. BadCrc = bad_crc_data(), ?m(?EXIT(data_error),(catch zlib:gunzip(BadCrc))), @@ -762,13 +801,13 @@ zip_usage({run,ZIP,ORIG}) -> ?m(ok, zlib:deflateInit(Z, default, deflated, -15, 8, default)), C2 = zlib:deflate(Z, ORIG, finish), - ?m(true, C1 == list_to_binary(C2)), + ?m(ORIG, zlib:unzip(C2)), ?m(ok, zlib:deflateEnd(Z)), ?m(ok, zlib:deflateInit(Z, none, deflated, -15, 8, filtered)), ?m(ok, zlib:deflateParams(Z, default, default)), C3 = zlib:deflate(Z, ORIG, finish), - ?m(true, C1 == list_to_binary(C3)), + ?m(ORIG, zlib:unzip(C3)), ?m(ok, zlib:deflateEnd(Z)), ok = zlib:close(Z), diff --git a/lib/mnesia/src/mnesia.erl b/lib/mnesia/src/mnesia.erl index b68b2de028..190fb2b56d 100644 --- a/lib/mnesia/src/mnesia.erl +++ b/lib/mnesia/src/mnesia.erl @@ -2681,7 +2681,7 @@ del_table_index(Tab, Ix) -> -spec transform_table(Tab::table(), Fun, [Attr]) -> t_result(ok) when Attr :: atom(), - Fun:: fun((Record::tuple()) -> Transformed::tuple()). + Fun:: fun((Record::tuple()) -> Transformed::tuple()) | ignore. transform_table(Tab, Fun, NewA) -> try val({Tab, record_name}) of OldRN -> mnesia_schema:transform_table(Tab, Fun, NewA, OldRN) @@ -2692,7 +2692,7 @@ transform_table(Tab, Fun, NewA) -> -spec transform_table(Tab::table(), Fun, [Attr], RecName) -> t_result(ok) when RecName :: atom(), Attr :: atom(), - Fun:: fun((Record::tuple()) -> Transformed::tuple()). + Fun:: fun((Record::tuple()) -> Transformed::tuple()) | ignore. transform_table(Tab, Fun, NewA, NewRN) -> mnesia_schema:transform_table(Tab, Fun, NewA, NewRN). diff --git a/lib/observer/src/cdv_bin_cb.erl b/lib/observer/src/cdv_bin_cb.erl index 5502869973..a4a542297c 100644 --- a/lib/observer/src/cdv_bin_cb.erl +++ b/lib/observer/src/cdv_bin_cb.erl @@ -71,6 +71,8 @@ hex_binary_fun(Bin) -> plain_html(io_lib:format("~s",[S])) end. +format_hex(<<>>,_) -> + []; format_hex(<<B1:4,B2:4>>,_) -> [integer_to_list(B1,16),integer_to_list(B2,16)]; format_hex(<<B1:4,B2:4,Bin/binary>>,0) -> diff --git a/lib/observer/src/crashdump_viewer.erl b/lib/observer/src/crashdump_viewer.erl index 7f6bb9bdcd..50e9102ff0 100644 --- a/lib/observer/src/crashdump_viewer.erl +++ b/lib/observer/src/crashdump_viewer.erl @@ -26,10 +26,25 @@ %% Tables %% ------ %% cdv_dump_index_table: This table holds all tags read from the -%% crashdump. Each tag indicates where the information about a -%% specific item starts. The table entry for a tag includes the start -%% position for this item-information. In a crash dump file, all tags -%% start with a "=" at the beginning of a line. +%% crashdump, except the 'binary' tag. Each tag indicates where the +%% information about a specific item starts. The table entry for a +%% tag includes the start position for this item-information. In a +%% crash dump file, all tags start with a "=" at the beginning of a +%% line. +%% +%% cdv_binary_index_table: This table holds all 'binary' tags. The hex +%% address for each binary is converted to its integer value before +%% storing Address -> Start Position in this table. The hex value of +%% the address is never used for lookup. +%% +%% cdv_reg_proc_table: This table holds mappings between pid and +%% registered name. This is used for timers and monitors. +%% +%% cdv_heap_file_chars: For each 'proc_heap' and 'literals' tag, this +%% table contains the number of characters to read from the crash dump +%% file. This is used for giving an indication in percent of the +%% progress when parsing this data. +%% %% %% Process state %% ------------- @@ -73,6 +88,9 @@ -export([init/1, handle_call/3, handle_cast/2, handle_info/2, terminate/2, code_change/3]). +%% Test support +-export([get_dump_versions/0]). + %% Debug support -export([debug/1,stop_debug/0]). @@ -87,6 +105,7 @@ % line_head/1 function can return -define(not_available,"N/A"). -define(binary_size_progress_limit,10000). +-define(max_dump_version,[0,4]). %% All possible tags - use macros in order to avoid misspelling in the code @@ -294,6 +313,11 @@ port(Id) -> expand_binary(Pos) -> call({expand_binary,Pos}). +%%%----------------------------------------------------------------- +%%% For testing only - called from crashdump_viewer_SUITE +get_dump_versions() -> + call(get_dump_versions). + %%==================================================================== %% Server functions %%==================================================================== @@ -455,8 +479,9 @@ handle_call(index_tables,_From,State=#state{file=File}) -> handle_call(schedulers,_From,State=#state{file=File}) -> Schedulers=schedulers(File), TW = truncated_warning([?scheduler]), - {reply,{ok,Schedulers,TW},State}. - + {reply,{ok,Schedulers,TW},State}; +handle_call(get_dump_versions,_From,State=#state{dump_vsn=DumpVsn}) -> + {reply,{ok,{?max_dump_version,DumpVsn}},State}. %%-------------------------------------------------------------------- @@ -781,10 +806,8 @@ parse_vsn_str(Str,WS) -> %%%----------------------------------------------------------------- -%%% Traverse crash dump and insert index in table for each heading -%%% -%%% Progress is reported during the time and MUST be checked with -%%% crashdump_viewer:get_progress/0 until it returns {ok,done}. +%%% Traverse crash dump and insert index in table for each heading. +%%% Progress is reported during the time. do_read_file(File) -> erase(?literals), %Clear literal cache. put(truncated,false), %Not truncated (yet). @@ -800,17 +823,21 @@ do_read_file(File) -> {Tag,Id,Rest,N1} = tag(Fd,TagAndRest,1), case Tag of ?erl_crash_dump -> - reset_tables(), - insert_index(Tag,Id,N1+1), - put_last_tag(Tag,""), - DumpVsn = [list_to_integer(L) || - L<-string:lexemes(Id,".")], - AddrAdj = get_bin_addr_adj(DumpVsn), - indexify(Fd,AddrAdj,Rest,N1), - end_progress(), - check_if_truncated(), - close(Fd), - {ok,DumpVsn}; + case check_dump_version(Id) of + {ok,DumpVsn} -> + reset_tables(), + insert_index(Tag,Id,N1+1), + put_last_tag(Tag,""), + AddrAdj = get_bin_addr_adj(DumpVsn), + indexify(Fd,AddrAdj,Rest,N1), + end_progress(), + check_if_truncated(), + close(Fd), + {ok,DumpVsn}; + Error -> + close(Fd), + Error + end; _Other -> R = io_lib:format( "~ts is not an Erlang crash dump~n", @@ -838,6 +865,18 @@ do_read_file(File) -> {error,R} end. +check_dump_version(Vsn) -> + DumpVsn = [list_to_integer(L) || L<-string:lexemes(Vsn,".")], + if DumpVsn > ?max_dump_version -> + Info = + "This Crashdump Viewer is too old for the given " + "Erlang crash dump. Please use a newer version of " + "Crashdump Viewer.", + {error,Info}; + true -> + {ok,DumpVsn} + end. + indexify(Fd,AddrAdj,Bin,N) -> case binary:match(Bin,<<"\n=">>) of {Start,Len} -> @@ -1083,7 +1122,7 @@ get_proc_details(File,Pid,WS,DumpVsn) -> {{Stack,MsgQ,Dict},TW} = case truncated_warning([{?proc,Pid}]) of [] -> - {expand_memory(Fd,Pid,DumpVsn),[]}; + expand_memory(Fd,Pid,DumpVsn); TW0 -> {{[],[],[]},TW0} end, @@ -1418,7 +1457,15 @@ expand_memory(Fd,Pid,DumpVsn) -> read_messages(Fd,Pid,BinAddrAdj,Dict), read_dictionary(Fd,Pid,BinAddrAdj,Dict)}, erase(fd), - Expanded. + IncompleteWarning = + case erase(incomplete_heap) of + undefined -> + []; + true -> + ["WARNING: This process has an incomplete heap. " + "Some information might be missing."] + end, + {Expanded,IncompleteWarning}. read_literals(Fd) -> case lookup_index(?literals,[]) of @@ -2747,6 +2794,7 @@ deref_ptr(Ptr, Line, BinAddrAdj, D0) -> none -> case get(fd) of end_of_heap -> + put(incomplete_heap,true), {['#CDVIncompleteHeap'],Line,D0}; Fd -> case bytes(Fd) of diff --git a/lib/observer/src/observer_html_lib.erl b/lib/observer/src/observer_html_lib.erl index 68381bcc7b..8202f91030 100644 --- a/lib/observer/src/observer_html_lib.erl +++ b/lib/observer/src/observer_html_lib.erl @@ -355,11 +355,11 @@ href_proc_bin(From, T, Acc, LTB) -> PreviewStr end end; - [PreviewIntStr,SizeStr,Md5] when From =:= obs -> + [PreviewIntStr,PreviewBitSizeStr,SizeStr,Md5] when From =:= obs -> Size = list_to_integer(SizeStr), PreviewInt = list_to_integer(PreviewIntStr), - PrevSize = (trunc(math:log2(PreviewInt)/8)+1)*8, - PreviewStr = preview_string(Size,<<PreviewInt:PrevSize>>), + PreviewBitSize = list_to_integer(PreviewBitSizeStr), + PreviewStr = preview_string(Size,<<PreviewInt:PreviewBitSize>>), if LTB -> href("TARGET=\"expanded\"", ["#OBSBinary?key1="++PreviewIntStr++ diff --git a/lib/observer/src/observer_lib.erl b/lib/observer/src/observer_lib.erl index e5ffe61d25..52e6c3e52a 100644 --- a/lib/observer/src/observer_lib.erl +++ b/lib/observer/src/observer_lib.erl @@ -810,7 +810,7 @@ progress_dialog_destroy({Dialog,_,_}) -> make_obsbin(Bin,Tab) -> Size = byte_size(Bin), - Preview = + {Preview,PreviewBitSize} = try %% The binary might be a unicode string, in which case we %% don't want to split it in the middle of a grapheme @@ -819,14 +819,14 @@ make_obsbin(Bin,Tab) -> PB1 = string:slice(Bin,0,PL1), PS1 = byte_size(PB1) * 8, <<P1:PS1>> = PB1, - P1 + {P1,PS1} catch _:_ -> %% Probably not a string, so just split anywhere PS2 = min(Size, 10) * 8, <<P2:PS2, _/binary>> = Bin, - P2 + {P2,PS2} end, Hash = erlang:phash2(Bin), Key = {Preview, Size, Hash}, ets:insert(Tab, {Key,Bin}), - ['#OBSBin',Preview,Size,Hash]. + ['#OBSBin',Preview,PreviewBitSize,Size,Hash]. diff --git a/lib/observer/test/crashdump_viewer_SUITE.erl b/lib/observer/test/crashdump_viewer_SUITE.erl index 86a60e15f4..29b9e406ae 100644 --- a/lib/observer/test/crashdump_viewer_SUITE.erl +++ b/lib/observer/test/crashdump_viewer_SUITE.erl @@ -25,7 +25,7 @@ %% Test functions -export([all/0, suite/0,groups/0,init_per_group/2,end_per_group/2, start_stop/1,load_file/1,not_found_items/1, - non_existing/1,not_a_crashdump/1,old_crashdump/1]). + non_existing/1,not_a_crashdump/1,old_crashdump/1,new_crashdump/1]). -export([init_per_suite/1, end_per_suite/1]). -export([init_per_testcase/2, end_per_testcase/2]). @@ -83,6 +83,7 @@ all() -> non_existing, not_a_crashdump, old_crashdump, + new_crashdump, load_file, not_found_items ]. @@ -212,6 +213,25 @@ not_a_crashdump(Config) when is_list(Config) -> ok = crashdump_viewer:stop(). +%% Try to load a file with newer version than this crashdump viewer can handle +new_crashdump(Config) -> + Dump = hd(?config(dumps,Config)), + ok = start_backend(Dump), + {ok,{MaxVsn,CurrentVsn}} = crashdump_viewer:get_dump_versions(), + if MaxVsn =/= CurrentVsn -> + ct:fail("Current dump version is not equal to cdv's max version"); + true -> + ok + end, + ok = crashdump_viewer:stop(), + NewerVsn = lists:join($.,[integer_to_list(X+1) || X <- MaxVsn]), + PrivDir = ?config(priv_dir,Config), + NewDump = filename:join(PrivDir,"new_erl_crash.dump"), + ok = file:write_file(NewDump,"=erl_crash_dump:"++NewerVsn++"\n"), + {error, Reason} = start_backend(NewDump), + "This Crashdump Viewer is too old" ++_ = Reason, + ok = crashdump_viewer:stop(). + %% Load files into the tool and view all pages load_file(Config) when is_list(Config) -> case ?t:is_debug() of @@ -328,7 +348,7 @@ browse_file(File) -> io:format(" info read",[]), - lookat_all_pids(Procs), + lookat_all_pids(Procs,is_truncated(File),incomplete_allowed(File)), io:format(" pids ok",[]), lookat_all_ports(Ports), io:format(" ports ok",[]), @@ -339,6 +359,21 @@ browse_file(File) -> Procs. % used as second arg to special/2 +is_truncated(File) -> + case filename:extension(File) of + ".trunc"++_ -> + true; + _ -> + false + end. + +incomplete_allowed(File) -> + %% Incomplete heap is allowed for native libs, since some literals + %% are not dumped - and for pre OTP-20 (really pre 20.2) releases, + %% since literals were not dumped at all then. + Rel = get_rel_from_dump_name(File), + Rel < 20 orelse test_server:is_native(lists). + special(File,Procs) -> case filename:extension(File) of ".full_dist" -> @@ -528,7 +563,7 @@ special(File,Procs) -> io:format(" process details ok",[]), #proc{dict=Dict} = ProcDetails, - io:format("~p\n", [Dict]), + %% io:format("~p\n", [Dict]), Maps = crashdump_helper:create_maps(), Maps = proplists:get_value(maps,Dict), io:format(" maps ok",[]), @@ -548,14 +583,25 @@ verify_binaries([Bin|T1], [['#CDVBin',Offset,Size,Pos]|T2]) -> verify_binaries([], []) -> ok. -lookat_all_pids([]) -> +lookat_all_pids([],_,_) -> ok; -lookat_all_pids([#proc{pid=Pid0}|Procs]) -> +lookat_all_pids([#proc{pid=Pid0}|Procs],TruncAllowed,IncompAllowed) -> Pid = pid_to_list(Pid0), - {ok,_ProcDetails=#proc{},_ProcTW} = crashdump_viewer:proc_details(Pid), - {ok,_Ets,_EtsTW} = crashdump_viewer:ets_tables(Pid), - {ok,_Timers,_TimersTW} = crashdump_viewer:timers(Pid), - lookat_all_pids(Procs). + {ok,_ProcDetails=#proc{},ProcTW} = crashdump_viewer:proc_details(Pid), + {ok,_Ets,EtsTW} = crashdump_viewer:ets_tables(Pid), + {ok,_Timers,TimersTW} = crashdump_viewer:timers(Pid), + case {ProcTW,EtsTW,TimersTW} of + {[],[],[]} -> + ok; + {["WARNING: This process has an incomplete heap."++_],[],[]} + when IncompAllowed -> + ok; % native libs, literals might not be included in dump + _ when TruncAllowed -> + ok; % truncated dump + TWs -> + ct:fail({unexpected_warning,TWs}) + end, + lookat_all_pids(Procs,TruncAllowed,IncompAllowed). lookat_all_ports([]) -> ok; @@ -603,13 +649,7 @@ do_create_dumps(DataDir,Rel) -> current -> CD3 = dump_with_args(DataDir,Rel,"instr","+Mim true"), CD4 = dump_with_strange_module_name(DataDir,Rel,"strangemodname"), - Tmp = dump_with_args(DataDir,Rel,"trunc_bytes",""), - {ok,#file_info{size=Max}} = file:read_file_info(Tmp), - ok = file:delete(Tmp), - Bytes = max(15,rand:uniform(Max)), - CD5 = dump_with_args(DataDir,Rel,"trunc_bytes", - "-env ERL_CRASH_DUMP_BYTES " ++ - integer_to_list(Bytes)), + CD5 = dump_with_size_limit_reached(DataDir,Rel,"trunc_bytes"), CD6 = dump_with_unicode_atoms(DataDir,Rel,"unicode"), CD7 = dump_with_maps(DataDir,Rel,"maps"), TruncatedDumps = truncate_dump(CD1), @@ -722,6 +762,28 @@ dump_with_strange_module_name(DataDir,Rel,DumpName) -> ?t:stop_node(n1), CD. +dump_with_size_limit_reached(DataDir,Rel,DumpName) -> + Tmp = dump_with_args(DataDir,Rel,DumpName,""), + {ok,#file_info{size=Max}} = file:read_file_info(Tmp), + ok = file:delete(Tmp), + dump_with_size_limit_reached(DataDir,Rel,DumpName,Max). + +dump_with_size_limit_reached(DataDir,Rel,DumpName,Max) -> + Bytes = max(15,rand:uniform(Max)), + CD = dump_with_args(DataDir,Rel,DumpName, + "-env ERL_CRASH_DUMP_BYTES " ++ + integer_to_list(Bytes)), + {ok,#file_info{size=Size}} = file:read_file_info(CD), + if Size < Bytes -> + %% This means that the dump was actually smaller than the + %% randomly selected truncation size, so we'll just do it + %% again with a smaller numer + ok = file:delete(CD), + dump_with_size_limit_reached(DataDir,Rel,DumpName,Size-3); + true -> + CD + end. + dump_with_unicode_atoms(DataDir,Rel,DumpName) -> Opt = rel_opt(Rel), Pz = "-pz \"" ++ filename:dirname(code:which(?MODULE)) ++ "\"", @@ -794,6 +856,11 @@ dump_prefix(current) -> dump_prefix(Rel) -> lists:concat(["r",Rel,"_dump."]). +get_rel_from_dump_name(File) -> + Name = filename:basename(File), + ["r"++Rel|_] = string:split(Name,"_"), + list_to_integer(Rel). + compat_rel(current) -> ""; compat_rel(Rel) -> diff --git a/lib/snmp/test/snmp_to_snmpnet_SUITE.erl b/lib/snmp/test/snmp_to_snmpnet_SUITE.erl index 24c14d86ea..6a3466b6e4 100644 --- a/lib/snmp/test/snmp_to_snmpnet_SUITE.erl +++ b/lib/snmp/test/snmp_to_snmpnet_SUITE.erl @@ -88,8 +88,17 @@ groups() -> ]. init_per_suite(Config) -> - [{agent_port, ?AGENT_PORT}, {manager_port, ?MANAGER_PORT} | Config]. - + case re:run(os:cmd("snmpd -v"),"NET-SNMP", [{capture, first}]) of + nomatch -> + {skip, "snmpd is NOT NET-SNMP"}; + {match, _} -> + case re:run(os:cmd("snmpd -v"),"5.4|5.6.2.1", [{capture, first}]) of + nomatch -> + [{agent_port, ?AGENT_PORT}, {manager_port, ?MANAGER_PORT} | Config]; + {match, _} -> + {skip, "buggy snmpd"} + end + end. end_per_suite(_Config) -> ok. @@ -322,7 +331,7 @@ snmpget(Oid, Transport, Config) -> Args = ["-c", "public", net_snmp_version(Versions), - "-m", "", + "-m", ":", "-Cf", net_snmp_addr_str(Transport), oid_str(Oid)], @@ -353,11 +362,13 @@ start_snmpd(Community, SysDescr, Config) -> ["--rocommunity"++domain_suffix(Domain)++"=" ++Community++" "++inet_parse:ntoa(Ip) || {Domain, {Ip, _}} <- Targets], + SnmpdArgs = - ["-f", "-r", %"-Dverbose", - "-c", filename:join(DataDir, "snmpd.conf"), - "-C", "-Lo", - "-m", "", + ["-f", "-r", %"-Dverbose", + "-c", filename:join(DataDir, "snmpd.conf"), + "-C", + "-Lo", + "-m", ":", "--sysDescr="++SysDescr, "--agentXSocket=tcp:localhost:"++integer_to_list(Port)] ++ CommunityArgs diff --git a/lib/ssh/doc/src/ssh_client_key_api.xml b/lib/ssh/doc/src/ssh_client_key_api.xml index a1cd9d4b02..98a1676ca4 100644 --- a/lib/ssh/doc/src/ssh_client_key_api.xml +++ b/lib/ssh/doc/src/ssh_client_key_api.xml @@ -56,11 +56,17 @@ <tag><c>string() =</c></tag> <item><p><c>[byte()]</c></p></item> <tag><c>public_key() =</c></tag> - <item><p><c>#'RSAPublicKey'{}| {integer(), #'Dss-Parms'{}}| term()</c></p></item> + <item><p><c>#'RSAPublicKey'{} + | {integer(),#'Dss-Parms'{}} + | {#'ECPoint'{},{namedCurve,Curve::string()}}</c></p></item> <tag><c>private_key() =</c></tag> - <item><p><c>#'RSAPrivateKey'{} | #'DSAPrivateKey'{} | term()</c></p></item> + <item><p><c>#'RSAPrivateKey'{} + | #'DSAPrivateKey'{} + | #'ECPrivateKey'{}</c></p></item> <tag><c>public_key_algorithm() =</c></tag> - <item><p><c>'ssh-rsa'| 'ssh-dss' | atom()</c></p></item> + <item><p><c>'ssh-rsa' | 'ssh-dss' + | 'rsa-sha2-256' | 'rsa-sha2-384' | 'rsa-sha2-512' + | 'ecdsa-sha2-nistp256' | 'ecdsa-sha2-nistp384' | 'ecdsa-sha2-nistp521' </c></p></item> </taglist> </section> @@ -73,10 +79,11 @@ <d>Description of the host that owns the <c>PublicKey</c>.</d> <v>Key = public_key()</v> - <d>Normally an RSA or DSA public key, but handling of other public keys can be added.</d> + <d>Normally an RSA, DSA or ECDSA public key, but handling of other public keys can be added.</d> <v>ConnectOptions = proplists:proplist()</v> - <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso></d> + <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso>. The option list given in + the <c>key_cb</c> option is available with the key <c>key_cb_private</c>.</d> <v>Reason = term().</v> </type> <desc> @@ -89,17 +96,17 @@ <fsummary>Checks if a host key is trusted.</fsummary> <type> <v>Key = public_key() </v> - <d>Normally an RSA or DSA public key, but handling of other public keys can be added.</d> + <d>Normally an RSA, DSA or ECDSA public key, but handling of other public keys can be added.</d> <v>Host = string()</v> <d>Description of the host.</d> <v>Algorithm = public_key_algorithm()</v> - <d>Host key algorithm. Is to support <c>'ssh-rsa'| 'ssh-dss'</c>, but more algorithms - can be handled.</d> + <d>Host key algorithm.</d> <v>ConnectOptions = proplists:proplist() </v> - <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso>.</d> + <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso>. The option list given in + the <c>key_cb</c> option is available with the key <c>key_cb_private</c>.</d> <v>Result = boolean()</v> </type> @@ -110,15 +117,15 @@ <func> <name>Module:user_key(Algorithm, ConnectOptions) -> - {ok, PrivateKey} | {error, Reason}</name> + {ok, PrivateKey} | {error, Reason}</name> <fsummary>Fetches the users <em>public key</em> matching the <c>Algorithm</c>.</fsummary> <type> <v>Algorithm = public_key_algorithm()</v> - <d>Host key algorithm. Is to support <c>'ssh-rsa'| 'ssh-dss'</c> but more algorithms - can be handled.</d> + <d>Host key algorithm.</d> <v>ConnectOptions = proplists:proplist()</v> - <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso></d> + <d>Options provided to <seealso marker="ssh#connect-3">ssh:connect/[3,4]</seealso>. The option list given in + the <c>key_cb</c> option is available with the key <c>key_cb_private</c>.</d> <v>PrivateKey = private_key()</v> <d>Private key of the user matching the <c>Algorithm</c>.</d> diff --git a/lib/ssh/doc/src/ssh_server_key_api.xml b/lib/ssh/doc/src/ssh_server_key_api.xml index a0694ca8d9..c6808b95d1 100644 --- a/lib/ssh/doc/src/ssh_server_key_api.xml +++ b/lib/ssh/doc/src/ssh_server_key_api.xml @@ -57,11 +57,17 @@ <tag><c>string() =</c></tag> <item><p><c>[byte()]</c></p></item> <tag><c>public_key() =</c></tag> - <item><p><c>#'RSAPublicKey'{}| {integer(), #'Dss-Parms'{}}| term()</c></p></item> + <item><p><c>#'RSAPublicKey'{} + | {integer(),#'Dss-Parms'{}} + | {#'ECPoint'{},{namedCurve,Curve::string()}}</c></p></item> <tag><c>private_key() =</c></tag> - <item><p><c>#'RSAPrivateKey'{} | #'DSAPrivateKey'{} | term()</c></p></item> + <item><p><c>#'RSAPrivateKey'{} + | #'DSAPrivateKey'{} + | #'ECPrivateKey'{}</c></p></item> <tag><c>public_key_algorithm() =</c></tag> - <item><p><c>'ssh-rsa'| 'ssh-dss' | atom()</c></p></item> + <item><p><c>'ssh-rsa' | 'ssh-dss' + | 'rsa-sha2-256' | 'rsa-sha2-384' | 'rsa-sha2-512' + | 'ecdsa-sha2-nistp256' | 'ecdsa-sha2-nistp384' | 'ecdsa-sha2-nistp521' </c></p></item> </taglist> </section> @@ -72,12 +78,13 @@ <fsummary>Fetches the host’s private key.</fsummary> <type> <v>Algorithm = public_key_algorithm()</v> - <d>Host key algorithm. Is to support <c>'ssh-rsa' | 'ssh-dss'</c>, but more algorithms - can be handled.</d> + <d>Host key algorithm.</d> <v>DaemonOptions = proplists:proplist()</v> - <d>Options provided to <seealso marker="ssh#daemon-2">ssh:daemon/[2,3]</seealso>.</d> - <v>Key = private_key()</v> - <d>Private key of the host matching the <c>Algorithm</c>.</d> + <d>Options provided to <seealso marker="ssh#daemon-2">ssh:daemon/[2,3]</seealso>. The option list given in + the <c>key_cb</c> option is available with the key <c>key_cb_private</c>.</d> + <v>Key = private_key() | crypto:engine_key_ref()</v> + <d>Private key of the host matching the <c>Algorithm</c>. + It may be a reference to a 'ssh-rsa', rsa-sha2-* or 'ssh-dss' (NOT ecdsa) key stored in a loaded Engine.</d> <v>Reason = term()</v> </type> <desc> @@ -90,11 +97,12 @@ <fsummary>Checks if the user key is authorized.</fsummary> <type> <v>Key = public_key()</v> - <d>Normally an RSA or DSA public key, but handling of other public keys can be added</d> + <d>Normally an RSA, DSA or ECDSA public key, but handling of other public keys can be added</d> <v>User = string()</v> <d>User owning the public key.</d> <v>DaemonOptions = proplists:proplist()</v> - <d>Options provided to <seealso marker="ssh#daemon-2">ssh:daemon/[2,3]</seealso>.</d> + <d>Options provided to <seealso marker="ssh#daemon-2">ssh:daemon/[2,3]</seealso>. The option list given in + the <c>key_cb</c> option is available with the key <c>key_cb_private</c>.</d> <v>Result = boolean()</v> </type> <desc> diff --git a/lib/ssh/src/ssh_transport.erl b/lib/ssh/src/ssh_transport.erl index d8f7a96c15..892db6b64f 100644 --- a/lib/ssh/src/ssh_transport.erl +++ b/lib/ssh/src/ssh_transport.erl @@ -811,7 +811,15 @@ extract_public_key(#'DSAPrivateKey'{y = Y, p = P, q = Q, g = G}) -> {Y, #'Dss-Parms'{p=P, q=Q, g=G}}; extract_public_key(#'ECPrivateKey'{parameters = {namedCurve,OID}, publicKey = Q}) -> - {#'ECPoint'{point=Q}, {namedCurve,OID}}. + {#'ECPoint'{point=Q}, {namedCurve,OID}}; +extract_public_key(#{engine:=_, key_id:=_, algorithm:=Alg} = M) -> + case {Alg, crypto:privkey_to_pubkey(Alg, M)} of + {rsa, [E,N]} -> + #'RSAPublicKey'{modulus = N, publicExponent = E}; + {dss, [P,Q,G,Y]} -> + {Y, #'Dss-Parms'{p=P, q=Q, g=G}} + end. + verify_host_key(#ssh{algorithms=Alg}=SSH, PublicKey, Digest, {AlgStr,Signature}) -> @@ -1261,10 +1269,12 @@ payload(<<PacketLen:32, PaddingLen:8, PayloadAndPadding/binary>>) -> <<Payload:PayloadLen/binary, _/binary>> = PayloadAndPadding, Payload. +sign(SigData, HashAlg, #{algorithm:=dss} = Key) -> + mk_dss_sig(crypto:sign(dss, HashAlg, SigData, Key)); +sign(SigData, HashAlg, #{algorithm:=SigAlg} = Key) -> + crypto:sign(SigAlg, HashAlg, SigData, Key); sign(SigData, HashAlg, #'DSAPrivateKey'{} = Key) -> - DerSignature = public_key:sign(SigData, HashAlg, Key), - #'Dss-Sig-Value'{r = R, s = S} = public_key:der_decode('Dss-Sig-Value', DerSignature), - <<R:160/big-unsigned-integer, S:160/big-unsigned-integer>>; + mk_dss_sig(public_key:sign(SigData, HashAlg, Key)); sign(SigData, HashAlg, Key = #'ECPrivateKey'{}) -> DerEncodedSign = public_key:sign(SigData, HashAlg, Key), #'ECDSA-Sig-Value'{r=R, s=S} = public_key:der_decode('ECDSA-Sig-Value', DerEncodedSign), @@ -1272,6 +1282,12 @@ sign(SigData, HashAlg, Key = #'ECPrivateKey'{}) -> sign(SigData, HashAlg, Key) -> public_key:sign(SigData, HashAlg, Key). + +mk_dss_sig(DerSignature) -> + #'Dss-Sig-Value'{r = R, s = S} = public_key:der_decode('Dss-Sig-Value', DerSignature), + <<R:160/big-unsigned-integer, S:160/big-unsigned-integer>>. + + verify(PlainText, HashAlg, Sig, {_, #'Dss-Parms'{}} = Key) -> case Sig of <<R:160/big-unsigned-integer, S:160/big-unsigned-integer>> -> @@ -1823,6 +1839,8 @@ kex_alg_dependent({Min, NBits, Max, Prime, Gen, E, F, K}) -> %%%---------------------------------------------------------------- +valid_key_sha_alg(#{engine:=_, key_id:=_}, _Alg) -> true; % Engine key + valid_key_sha_alg(#'RSAPublicKey'{}, 'rsa-sha2-512') -> true; valid_key_sha_alg(#'RSAPublicKey'{}, 'rsa-sha2-384') -> true; valid_key_sha_alg(#'RSAPublicKey'{}, 'rsa-sha2-256') -> true; diff --git a/lib/ssh/test/Makefile b/lib/ssh/test/Makefile index 5ea048a352..a18383d148 100644 --- a/lib/ssh/test/Makefile +++ b/lib/ssh/test/Makefile @@ -38,6 +38,7 @@ MODULES= \ ssh_basic_SUITE \ ssh_bench_SUITE \ ssh_connection_SUITE \ + ssh_engine_SUITE \ ssh_protocol_SUITE \ ssh_property_test_SUITE \ ssh_sftp_SUITE \ @@ -49,6 +50,7 @@ MODULES= \ ssh_test_lib \ ssh_key_cb \ ssh_key_cb_options \ + ssh_key_cb_engine_keys \ ssh_trpt_test_lib \ ssh_echo_server \ ssh_bench_dev_null \ diff --git a/lib/ssh/test/ssh_engine_SUITE.erl b/lib/ssh/test/ssh_engine_SUITE.erl new file mode 100644 index 0000000000..035446932b --- /dev/null +++ b/lib/ssh/test/ssh_engine_SUITE.erl @@ -0,0 +1,141 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2008-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% + +%% + +-module(ssh_engine_SUITE). + +-include_lib("common_test/include/ct.hrl"). +-include("ssh_test_lib.hrl"). + +%% Note: This directive should only be used in test suites. +-compile(export_all). + +%%-------------------------------------------------------------------- +%% Common Test interface functions ----------------------------------- +%%-------------------------------------------------------------------- + +suite() -> + [{ct_hooks,[ts_install_cth]}, + {timetrap,{seconds,40}}]. + +all() -> + [{group, dsa_key}, + {group, rsa_key} + ]. + +groups() -> + [{dsa_key, [], basic_tests()}, + {rsa_key, [], basic_tests()} + ]. + +basic_tests() -> + [simple_connect + ]. + + +%%-------------------------------------------------------------------- +init_per_suite(Config) -> + ssh:start(), + ?CHECK_CRYPTO( + case load_engine() of + {ok,E} -> + ssh_dbg:messages(fun ct:pal/2), + [{engine,E}|Config]; + {error, notsup} -> + {skip, "Engine not supported on this OpenSSL version"}; + {error, bad_engine_id} -> + {skip, "Dynamic Engine not supported"}; + Other -> + ct:log("Engine load failed: ~p",[Other]), + {fail, "Engine load failed"} + end + ). + +end_per_suite(Config) -> + catch crypto:engine_unload( proplists:get_value(engine,Config) ), + ssh:stop(). + +%%-------------------------------------------------------------------- +init_per_group(dsa_key, Config) -> + case lists:member('ssh-dss', + ssh_transport:default_algorithms(public_key)) of + true -> + start_daemon(Config, 'ssh-dss', "dsa_private_key.pem"); + false -> + {skip, unsupported_pub_key} + end; +init_per_group(rsa_key, Config) -> + case lists:member('ssh-rsa', + ssh_transport:default_algorithms(public_key)) of + true -> + start_daemon(Config, 'ssh-rsa', "rsa_private_key.pem"); + false -> + {skip, unsupported_pub_key} + end. + +start_daemon(Config, KeyType, KeyId) -> + SystemDir = proplists:get_value(data_dir, Config), + FullKeyId = filename:join(SystemDir, KeyId), + KeyCBOpts = [{engine, proplists:get_value(engine,Config)}, + {KeyType, FullKeyId} + ], + Opts = [{key_cb, {ssh_key_cb_engine_keys, KeyCBOpts}}], + {Pid, Host, Port} = ssh_test_lib:std_daemon(Config, Opts), + [{host_port,{Host,Port}}, {daemon_pid,Pid}| Config]. + + +end_per_group(_, Config) -> + catch ssh:stop_daemon(proplists:get_value(daemon_pid,Config)), + Config. + +%%-------------------------------------------------------------------- +%% Test Cases -------------------------------------------------------- +%%-------------------------------------------------------------------- + +%% A simple exec call +simple_connect(Config) -> + {Host,Port} = proplists:get_value(host_port, Config), + CRef = ssh_test_lib:std_connect(Config, Host, Port, []), + ssh:close(CRef). + +%%-------------------------------------------------------------------- +%%-------------------------------------------------------------------- +load_engine() -> + case crypto:get_test_engine() of + {ok, Engine} -> + try crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, Engine}, + <<"LOAD">>], + []) + catch + error:notsup -> + {error, notsup} + end; + + {error, Error} -> + {error, Error} + end. + +start_std_daemon(Opts, Config) -> + ct:log("starting std_daemon",[]), + {Pid, Host, Port} = ssh_test_lib:std_daemon(Config, Opts), + ct:log("started ~p:~p ~p",[Host,Port,Opts]), + [{srvr_pid,Pid},{srvr_addr,{Host,Port}} | Config]. diff --git a/lib/ssh/test/ssh_engine_SUITE_data/dsa_private_key.pem b/lib/ssh/test/ssh_engine_SUITE_data/dsa_private_key.pem new file mode 100644 index 0000000000..778ffac675 --- /dev/null +++ b/lib/ssh/test/ssh_engine_SUITE_data/dsa_private_key.pem @@ -0,0 +1,9 @@ +-----BEGIN PRIVATE KEY----- +MIIBSwIBADCCASwGByqGSM44BAEwggEfAoGBAMyitTMR7vPbpqyAXJpqnB0AhFwQ +F87IE+JKFl5bD/MSkhhRV5sM73HUU1ooXY0FjhZ+cdLUCATuZR5ta4ydANqWIcAB +gX3IwF1B4zf5SXEKTWkUYneL9dOKtiZLtoG28swrk8xMxwX+0fLHkltCEj6FiTW9 +PFrv8GmIfV6DjcI9AhUAqXWbb3RtoN9Ld28fVMhGZrj3LJUCgYEAwnxGHGBMpJaF +2w7zAw3jHjL8PMYlV6vnufGHQlwF0ZUXJxRsvagMb/X1qACTu2VPYEVoLQGM3cfH +EhHoQmvSXGAyTfR7Bmn3gf1n/s/DcFbdZduUCZ/rAyIrfd0eSbc1I+kZk85UCsKK +w/IYdlqcuYa4Cgm2TapT5uEMqH4jhzEEFgIULh8swEUWmU8aJNWsrWl4eCiuUUg= +-----END PRIVATE KEY----- diff --git a/lib/ssh/test/ssh_engine_SUITE_data/ecdsa_private_key.pem b/lib/ssh/test/ssh_engine_SUITE_data/ecdsa_private_key.pem new file mode 100644 index 0000000000..a45522064f --- /dev/null +++ b/lib/ssh/test/ssh_engine_SUITE_data/ecdsa_private_key.pem @@ -0,0 +1,8 @@ +-----BEGIN PRIVATE KEY----- +MIHuAgEAMBAGByqGSM49AgEGBSuBBAAjBIHWMIHTAgEBBEIBparGjr0KcdNrVM2J +G0mW5ltP1QyvxDqBMyWLWo3fruRZv6Qoohl5skd1u4O+KJoM/UrrSTOXI/MDR7NN +i1yl7O+hgYkDgYYABAG8K2XVsK0ahG9+HIIPwCO0pJY8ulwSTXwIjkCGyB2lpglh +8qJmRzuyGcfRTslv8wfv0sPlT9H9PKDvgrTUL7rvQQDdOODNgVPXSecUoXoPn+X+ +eqxs77bjx+A5x0t/i3m5PfkaNPh5MZ1H/bWuOOdj2ZXZw0R4rlVc0zVrgnPU8L8S +BQ== +-----END PRIVATE KEY----- diff --git a/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key.pem b/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key.pem new file mode 100644 index 0000000000..ea0e3d3958 --- /dev/null +++ b/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key.pem @@ -0,0 +1,28 @@ +-----BEGIN PRIVATE KEY----- +MIIEvwIBADANBgkqhkiG9w0BAQEFAASCBKkwggSlAgEAAoIBAQCwwb0/ddXGXTFK +4FLxXdV6a/WJMSoPPS55RvZIAHFsiTtvPLbJ8LxDsZ6wSVZLN0/UQ4wdWn9jftyj +U5/IxBVG8XOtKimTMvm3/ZOzVLueGHBbrLYscRv9oL85ulTKHWgrZDu0lBX5JJTI +v5UTCErzJRQbka9DG1GaBgDb1PlXfkzBWMwfsBZmwoC77KvCcIGCgbW/XCY03TP2 +3Tg8drvpByMStddP2FQ4fZ91qFUzPu8uhZEsqSQTFlmhgGEx7dLlky0xvu62RuAD +RTpINpcWZtWDHTdssOqu653LwwqBY8lBopCZ/4Af8QR3ZYkQhen1YLEbVheXRuzI +LSCZIiJNAgMBAAECggEBAJH4/fxpqQkvr2Shy33Pu1xlyhnpw01gfn/jrcKasxEq +aC4eWup86E2TY3U8q4pkfIXU3uLi+O9HNpmflwargNLc1mY8uqb44ygiv5bLNEKE +9k2PXcdoBfC4jxPyoNFl5cBn/7LK1TazEjiTl15na9ZPWcLG1pG5/vMPYCgsQ1sP +8J3c4E3aaXIj9QceYxBprl490OCzieGyZlRipncz3g4UShRc/b4cycvDZOJpmAy4 +zbWTcBcSMPVPi5coF0K8UcimiqZkotfb/2RLc433i34IdsIXMM+brdq+g8rmjg5a ++oQPy02M6tFApBruEhAz8DGgaLtDY6MLtyZAt3SjXnUCgYEA1zLgamdTHOqrrmIi +eIQBnAJiyIfcY8B9SX1OsLGYFCHiPVwgUY35B2c7MavMsGcExJhtE+uxU7o5djtM +R6r9cRHOXJ6EQwa8OwzzPqbM17/YqNDeK39bc9WOFUqRWrhDhVMPy6z8rmZr73mG +IUC7mBNx/1GBdVYXIlsXzC96dI8CgYEA0kUAhz6I5nyPa70NDEUYHLHf3IW1BCmE +UoVbraSePJtIEY/IqFx7oDuFo30d4n5z+8ICCtyid1h/Cp3mf3akOiqltYUfgV1G +JgcEjKKYWEnO7cfFyO7LB7Y3GYYDJNy6EzVWPiwTGk9ZTfFJEESmHC45Unxgd17m +Dx/R58rFgWMCgYBQXQWFdtSI5fH7C1bIHrPjKNju/h2FeurOuObcAVZDnmu4cmD3 +U8d9xkVKxVeJQM99A1coq0nrdI3k4zwXP3mp8fZYjDHkPe2pN6rW6L9yiohEcsuk +/siON1/5/4DMmidM8LnjW9R45HLGWWGHpX7oyco2iJ+Jy/6Tq+T1MX3PbQKBgQCm +hdsbQJ0u3CrBSmFQ/E9SOlRt0r4+45pVuCOY6yweF2QF9HcXTtbhWQJHLclDHJ5C +Ha18aKuKFN3XzKFFBPKe1jOSBDGlQ/dQGnKx5fr8wMdObM3oiaTlIJuWbRmEUgJT +QARjDIi8Z2b0YUhZx+Q9oSXoe3PyVYehJrQX+/BavQKBgQCIr7Zp0rQPbfqcTL+M +OYHUoNcb14f9f8hXeXHQOqVpsGwxGdRQAU9wbx/4+obKB5xIkzBsVNcJwavisNja +hegnGjTB/9Hc4m+5bMGwH0bhS2eQO4o+YYM2ypDmFQqDLRfFUlZ5PVHffm/aA9+g +GanNBCsmtoHtV6CJ1UZ7NmBuIA== +-----END PRIVATE KEY----- diff --git a/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key_pwd.pem b/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key_pwd.pem new file mode 100644 index 0000000000..501662fc35 --- /dev/null +++ b/lib/ssh/test/ssh_engine_SUITE_data/rsa_private_key_pwd.pem @@ -0,0 +1,30 @@ +-----BEGIN ENCRYPTED PRIVATE KEY----- +MIIFDjBABgkqhkiG9w0BBQ0wMzAbBgkqhkiG9w0BBQwwDgQIh888Iq6gxuMCAggA +MBQGCCqGSIb3DQMHBAic/11YZ8Nt5gSCBMjG/Jb4qiMoBS50iQvHXqcETPE+0NBr +jhsn9w94LkdRBstMPAsoKmY98Er96Rnde/NfmqlU9CupKTkd7Ce5poBf72Y6KMED +cPURyjbGRFsu6x9skXB2obhyKYEqAEF2oQAg4Qbe5v1qXBIgDuC/NgiJnM+w2zCZ +LkHSZB2/NmcnvDzcgPF7TM8pTO23xCJ33m37qjfWvHsgocVqZmL9wQ4+wr/NMYjJ +pJvX1OHW1vBsZsXh40WchalYRSB1VeO368QfsE8coRJztqbMzdce9EQdMB6Q6jlO +cetd3moLIoMP4I7HW0/SgokbycTbRiYSvRyU1TGc2WbW6BrFZV24IckcnnVUFatf +6HKUcaYLG68dJcRgs5QMGkcmgVvlddENHFmHZlo0eym/xSiUl/AT8/5odscm6ML8 +wW5sneax+TF4J2eYmiN7yjAUCodXVTNYNDVKo6uUhntlymbM0o4UitVIbPIfTDHl +sxJAEZ7vpuPqeNMxUk6G6zipuEjqsVbnuFSBSZmgKiGYcifRPUmqqINa3DdS4WVx +xaPWdHbHVRD//ze3h/FsA+1lIE5q2kUE0xXseJA1ISog++kJp14XeaaL2j/tx3Ob +OsbcaOAD/IUw/ItDt9kn0qzfnar7sS0Wov8AmJQxHmH7Lm93jHTLM05yE0AR/eBr +Mig2ZdC+9OqVC+GPuBkRjSs8NpltQIDroz6EV9IMwPwXm0szSYoyoPLmlHJUdnLs +ZUef+au6hYkEJBrvuisagnq5eT/fCV3hsjD7yODebNU2CmBTo6X2PRx/xsBHRMWl +QkoM9PBdSCnKv6HpHl4pchuoqU2NpFjN0BCaad6aHfZSTnqgzK4bEh1oO6dI8/rB +/eh71JyFFG5J4xbpaqz5Su01V1iwU5leK5bDwqals4M4+ZGHGciou7qnXUmX2fJl +r6DlMUa/xy+A2ZG0NuZR05yk2oB3+KVNMgp6zFty3XaxwoNtc8GTLtLnBnIh2rlP +mE1+I65LRWwrNQalPeOAUrYuEzhyp2Df7a8Ykas5PUH7MGR/S0Ge/dLxtE2bJuK4 +znbLAsGhvo/SbNxYqIp6D4iDtd3va6yUGncy41paA/vTKFVvXZDrXcwJQYYCVOGT +OwdzNuozU8Dc7oxsd8oakfC46kvmVaOrGvZbm56PFfprcaL/Hslska5xxEni/eZe +WRxZbCBhAVqS1pn5zkDQVUe9uFlR/x39Qi01HIlKLBsjpSs6qQsFArMe8hgXmXLG +xP+dyVuOE18NzSewdEjeqSRKIM7Qi8EOjZsI4HdSRBY7bh9VhmaVXDZiCSf33TTE +3y8nimzQAeuGoYg6WqHmWWC2Qnpki2HlaIH/ayXEyQWkP/qvg61e8ovdg9Fy8JOO +0AacXVt5zj0q00AW5bKx7usi4NIjZedi86hUm6H19aBm7r86BKjwYTEI/GOcdrbV +9HC/8ayOimgwiAG3gq+aLioWym+Z6KnsbVd7XReVbvM/InQx54WA2y5im0A+/c67 +oQFFPV84XGX9waeqv/K4Wzkm6HW+qVAEM67482VGOf0PVrlQMno6dOotT/Y7ljoZ +2iz0LmN9yylJnLPDrr1i6gzbs5OhhUgbF5LI2YP2wWdCZTl/DrKSIvQZWl8U+tw3 +ciA= +-----END ENCRYPTED PRIVATE KEY----- diff --git a/lib/ssh/test/ssh_key_cb_engine_keys.erl b/lib/ssh/test/ssh_key_cb_engine_keys.erl new file mode 100644 index 0000000000..fc9cbfd49b --- /dev/null +++ b/lib/ssh/test/ssh_key_cb_engine_keys.erl @@ -0,0 +1,62 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2015-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% + +%% +%%---------------------------------------------------------------------- + +%% Note: This module is used by ssh_basic_SUITE + +-module(ssh_key_cb_engine_keys). +-behaviour(ssh_server_key_api). +-compile(export_all). + +host_key(SshAlg, Options) -> + KBopts = proplists:get_value(key_cb_private, Options, []), + Engine = proplists:get_value(engine, KBopts), + case proplists:get_value(SshAlg, KBopts) of + undefined -> + {error, {unknown_alg,SshAlg}}; + KeyId -> + case crypto_alg(SshAlg) of + undefined -> + {error, {unsupported_alg,SshAlg}}; + CryptoAlg -> + PrivKey = #{engine => Engine, + key_id => KeyId, + algorithm => CryptoAlg}, + %% Is there a key with this reference ? + case crypto:privkey_to_pubkey(CryptoAlg, PrivKey) of + [_|_] -> + {ok, PrivKey}; + _ -> + {error, {no_hostkey,SshAlg}} + end + end + end. + +is_auth_key(_PublicUserKey, _User, _Options) -> + false. + + + +crypto_alg('ssh-rsa') -> rsa; +crypto_alg('ssh-dss') -> dss; +crypto_alg(_) -> undefined. + diff --git a/lib/ssl/doc/src/ssl.xml b/lib/ssl/doc/src/ssl.xml index e80fd59a7f..ac5a69c69b 100644 --- a/lib/ssl/doc/src/ssl.xml +++ b/lib/ssl/doc/src/ssl.xml @@ -69,7 +69,9 @@ <p><c>| {cert, public_key:der_encoded()}</c></p> <p><c>| {certfile, path()}</c></p> <p><c>| {key, {'RSAPrivateKey'| 'DSAPrivateKey' | 'ECPrivateKey' - | 'PrivateKeyInfo', public_key:der_encoded()}}</c></p> + | 'PrivateKeyInfo', public_key:der_encoded()} | + #{algorithm := rsa | dss | ecdsa, + engine := crypto:engine_ref(), key_id := crypto:key_id(), password => crypto:password()}</c></p> <p><c>| {keyfile, path()}</c></p> <p><c>| {password, string()}</c></p> <p><c>| {cacerts, [public_key:der_encoded()]}</c></p> @@ -202,8 +204,12 @@ <item><p>Path to a file containing the user certificate.</p></item> <tag><c>{key, {'RSAPrivateKey'| 'DSAPrivateKey' | 'ECPrivateKey' - |'PrivateKeyInfo', public_key:der_encoded()}}</c></tag> - <item><p>The DER-encoded user's private key. If this option + |'PrivateKeyInfo', public_key:der_encoded()} | #{algorithm := rsa | dss | ecdsa, + engine := crypto:engine_ref(), key_id := crypto:key_id(), password => crypto:password()}</c></tag> + <item><p>The DER-encoded user's private key or a map refering to a crypto + engine and its key reference that optionally can be password protected, + seealso <seealso marker="crypto:crypto#engine_load-4"> crypto:engine_load/4 + </seealso> and <seealso marker="crypto:engine_load"> Crypto's Users Guide</seealso>. If this option is supplied, it overrides option <c>keyfile</c>.</p></item> <tag><c>{keyfile, path()}</c></tag> diff --git a/lib/ssl/src/ssl.erl b/lib/ssl/src/ssl.erl index 60118549e4..4007e44a83 100644 --- a/lib/ssl/src/ssl.erl +++ b/lib/ssl/src/ssl.erl @@ -896,7 +896,8 @@ validate_option(key, {KeyType, Value}) when is_binary(Value), KeyType == 'ECPrivateKey'; KeyType == 'PrivateKeyInfo' -> {KeyType, Value}; - +validate_option(key, #{algorithm := _} = Value) -> + Value; validate_option(keyfile, undefined) -> <<>>; validate_option(keyfile, Value) when is_binary(Value) -> diff --git a/lib/ssl/src/ssl_config.erl b/lib/ssl/src/ssl_config.erl index e4611995ec..022fb7eac0 100644 --- a/lib/ssl/src/ssl_config.erl +++ b/lib/ssl/src/ssl_config.erl @@ -91,7 +91,15 @@ init_certificates(undefined, #{pem_cache := PemCache} = Config, CertFile, server end; init_certificates(Cert, Config, _, _) -> {ok, Config#{own_certificate => Cert}}. - +init_private_key(_, #{algorithm := Alg} = Key, <<>>, _Password, _Client) when Alg == ecdsa; + Alg == rsa; + Alg == dss -> + case maps:is_key(engine, Key) andalso maps:is_key(key_id, Key) of + true -> + Key; + false -> + throw({key, {invalid_key_id, Key}}) + end; init_private_key(_, undefined, <<>>, _Password, _Client) -> undefined; init_private_key(DbHandle, undefined, KeyFile, Password, _) -> diff --git a/lib/ssl/src/ssl_handshake.erl b/lib/ssl/src/ssl_handshake.erl index 81d38a38e4..8681765284 100644 --- a/lib/ssl/src/ssl_handshake.erl +++ b/lib/ssl/src/ssl_handshake.erl @@ -1702,14 +1702,20 @@ digitally_signed(Version, Hashes, HashAlgo, PrivateKey) -> error:badkey-> throw(?ALERT_REC(?FATAL, ?HANDSHAKE_FAILURE, bad_key(PrivateKey))) end. - +do_digitally_signed({3, Minor}, Hash, HashAlgo, #{algorithm := Alg} = Engine) + when Minor >= 3 -> + crypto:sign(Alg, HashAlgo, {digest, Hash}, maps:remove(algorithm, Engine)); do_digitally_signed({3, Minor}, Hash, HashAlgo, Key) when Minor >= 3 -> public_key:sign({digest, Hash}, HashAlgo, Key); -do_digitally_signed(_Version, Hash, HashAlgo, #'DSAPrivateKey'{} = Key) -> - public_key:sign({digest, Hash}, HashAlgo, Key); do_digitally_signed(_Version, Hash, _HashAlgo, #'RSAPrivateKey'{} = Key) -> public_key:encrypt_private(Hash, Key, [{rsa_pad, rsa_pkcs1_padding}]); +do_digitally_signed({3, _}, Hash, _, + #{algorithm := rsa} = Engine) -> + crypto:private_encrypt(rsa, Hash, maps:remove(algorithm, Engine), + rsa_pkcs1_padding); +do_digitally_signed({3, _}, Hash, HashAlgo, #{algorithm := Alg} = Engine) -> + crypto:sign(Alg, HashAlgo, {digest, Hash}, maps:remove(algorithm, Engine)); do_digitally_signed(_Version, Hash, HashAlgo, Key) -> public_key:sign({digest, Hash}, HashAlgo, Key). diff --git a/lib/ssl/src/ssl_internal.hrl b/lib/ssl/src/ssl_internal.hrl index 24ac34653e..9bb1cbaeb0 100644 --- a/lib/ssl/src/ssl_internal.hrl +++ b/lib/ssl/src/ssl_internal.hrl @@ -95,7 +95,8 @@ certfile :: binary(), cert :: public_key:der_encoded() | secret_printout() | 'undefined', keyfile :: binary(), - key :: {'RSAPrivateKey' | 'DSAPrivateKey' | 'ECPrivateKey' | 'PrivateKeyInfo', public_key:der_encoded()} | secret_printout() | 'undefined', + key :: {'RSAPrivateKey' | 'DSAPrivateKey' | 'ECPrivateKey' | 'PrivateKeyInfo', + public_key:der_encoded()} | key_map() | secret_printout() | 'undefined', password :: string() | secret_printout() | 'undefined', cacerts :: [public_key:der_encoded()] | secret_printout() | 'undefined', cacertfile :: binary(), @@ -164,7 +165,15 @@ connection_cb }). - +-type key_map() :: #{algorithm := rsa | dss | ecdsa, + %% engine and key_id ought to + %% be :=, but putting it in + %% the spec gives dialyzer warning + %% of correct code! + engine => crypto:engine_ref(), + key_id => crypto:key_id(), + password => crypto:password() + }. -type state_name() :: hello | abbreviated | certify | cipher | connection. -type gen_fsm_state_return() :: {next_state, state_name(), term()} | {next_state, state_name(), term(), timeout()} | diff --git a/lib/ssl/test/Makefile b/lib/ssl/test/Makefile index c7e2f402af..aa01552c39 100644 --- a/lib/ssl/test/Makefile +++ b/lib/ssl/test/Makefile @@ -44,6 +44,7 @@ MODULES = \ ssl_certificate_verify_SUITE\ ssl_crl_SUITE\ ssl_dist_SUITE \ + ssl_engine_SUITE\ ssl_handshake_SUITE \ ssl_npn_hello_SUITE \ ssl_npn_handshake_SUITE \ diff --git a/lib/ssl/test/ssl_engine_SUITE.erl b/lib/ssl/test/ssl_engine_SUITE.erl new file mode 100644 index 0000000000..bc221d35fd --- /dev/null +++ b/lib/ssl/test/ssl_engine_SUITE.erl @@ -0,0 +1,142 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2017-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% + +%% +-module(ssl_engine_SUITE). + +%% Note: This directive should only be used in test suites. +-compile(export_all). + +-include_lib("common_test/include/ct.hrl"). +-include_lib("public_key/include/public_key.hrl"). + +%%-------------------------------------------------------------------- +%% Common Test interface functions ----------------------------------- +%%-------------------------------------------------------------------- +all() -> + [ + private_key + ]. + +init_per_suite(Config) -> + catch crypto:stop(), + try crypto:start() of + ok -> + ssl_test_lib:clean_start(), + case crypto:get_test_engine() of + {ok, EngineName} -> + try crypto:engine_load(<<"dynamic">>, + [{<<"SO_PATH">>, EngineName}, + <<"LOAD">>], + []) of + {ok, Engine} -> + [{engine, Engine} |Config]; + {error, Reason} -> + ct:pal("Reason ~p", [Reason]), + {skip, "No dynamic engine support"} + catch error:notsup -> + {skip, "No engine support in OpenSSL"} + end; + {error, notexist} -> + {skip, "Test engine not found"} + end + catch _:_ -> + {skip, "Crypto did not start"} + end. + +end_per_suite(Config) -> + Engine = proplists:get_value(engine, Config), + crypto:engine_unload(Engine), + ssl:stop(), + application:stop(crypto). + + +init_per_testcase(_TestCase, Config) -> + ssl:stop(), + ssl:start(), + ssl_test_lib:ct_log_supported_protocol_versions(Config), + ct:timetrap({seconds, 10}), + Config. + +end_per_testcase(_TestCase, Config) -> + Config. + +%%-------------------------------------------------------------------- +%% Test Cases -------------------------------------------------------- +%%-------------------------------------------------------------------- + +private_key(Config) when is_list(Config) -> + ClientFileBase = filename:join([proplists:get_value(priv_dir, Config), "client_engine"]), + ServerFileBase = filename:join([proplists:get_value(priv_dir, Config), "server_engine"]), + #{server_config := ServerConf, + client_config := ClientConf} = GenCertData = + public_key:pkix_test_data(#{server_chain => + #{root => [{key, ssl_test_lib:hardcode_rsa_key(1)}], + intermediates => [[{key, ssl_test_lib:hardcode_rsa_key(2)}]], + peer => [{key, ssl_test_lib:hardcode_rsa_key(3)} + ]}, + client_chain => + #{root => [{key, ssl_test_lib:hardcode_rsa_key(4)}], + intermediates => [[{key, ssl_test_lib:hardcode_rsa_key(5)}]], + peer => [{key, ssl_test_lib:hardcode_rsa_key(6)}]}}), + [{server_config, FileServerConf}, + {client_config, FileClientConf}] = + x509_test:gen_pem_config_files(GenCertData, ClientFileBase, ServerFileBase), + + Engine = proplists:get_value(engine, Config), + + ClientKey = engine_key(FileClientConf), + ServerKey = engine_key(FileServerConf), + + EngineClientConf = [{key, #{algorithm => rsa, + engine => Engine, + key_id => ClientKey}} | proplists:delete(key, ClientConf)], + + EngineServerConf = [{key, #{algorithm => rsa, + engine => Engine, + key_id => ServerKey}} | proplists:delete(key, ServerConf)], + %% Test with engine + test_tls_connection(EngineServerConf, EngineClientConf, Config), + %% Test that sofware fallback is available + test_tls_connection(ServerConf, [{reuse_sessions, false} |ClientConf], Config). + +engine_key(Conf) -> + FileStr = proplists:get_value(keyfile, Conf), + list_to_binary(FileStr). + + +test_tls_connection(ServerConf, ClientConf, Config) -> + {ClientNode, ServerNode, Hostname} = ssl_test_lib:run_where(Config), + + Server = ssl_test_lib:start_server([{node, ServerNode}, {port, 0}, + {from, self()}, + {mfa, {ssl_test_lib, send_recv_result_active, []}}, + {options, [{verify, verify_peer} + | ServerConf]}]), + Port = ssl_test_lib:inet_port(Server), + Client = ssl_test_lib:start_client([{node, ClientNode}, {port, Port}, + {host, Hostname}, + {from, self()}, + {mfa, {ssl_test_lib, send_recv_result_active, []}}, + {options, [{verify, verify_peer} | ClientConf]}]), + + ssl_test_lib:check_result(Server, ok, Client, ok), + ssl_test_lib:close(Server), + ssl_test_lib:close(Client). diff --git a/lib/stdlib/doc/src/Makefile b/lib/stdlib/doc/src/Makefile index 93eac8220d..aeed79408b 100644 --- a/lib/stdlib/doc/src/Makefile +++ b/lib/stdlib/doc/src/Makefile @@ -98,6 +98,7 @@ XML_REF3_FILES = \ sys.xml \ timer.xml \ unicode.xml \ + uri_string.xml \ win32reg.xml \ zip.xml diff --git a/lib/stdlib/doc/src/ref_man.xml b/lib/stdlib/doc/src/ref_man.xml index 878a3babc5..68bfddbc71 100644 --- a/lib/stdlib/doc/src/ref_man.xml +++ b/lib/stdlib/doc/src/ref_man.xml @@ -93,6 +93,7 @@ <xi:include href="sys.xml"/> <xi:include href="timer.xml"/> <xi:include href="unicode.xml"/> + <xi:include href="uri_string.xml"/> <xi:include href="win32reg.xml"/> <xi:include href="zip.xml"/> </application> diff --git a/lib/stdlib/doc/src/specs.xml b/lib/stdlib/doc/src/specs.xml index 45b207b13d..d559adf9b6 100644 --- a/lib/stdlib/doc/src/specs.xml +++ b/lib/stdlib/doc/src/specs.xml @@ -60,6 +60,7 @@ <xi:include href="../specs/specs_sys.xml"/> <xi:include href="../specs/specs_timer.xml"/> <xi:include href="../specs/specs_unicode.xml"/> + <xi:include href="../specs/specs_uri_string.xml"/> <xi:include href="../specs/specs_win32reg.xml"/> <xi:include href="../specs/specs_zip.xml"/> </specs> diff --git a/lib/stdlib/doc/src/uri_string.xml b/lib/stdlib/doc/src/uri_string.xml new file mode 100644 index 0000000000..9ace2b0a05 --- /dev/null +++ b/lib/stdlib/doc/src/uri_string.xml @@ -0,0 +1,230 @@ +<?xml version="1.0" encoding="utf-8" ?> +<!DOCTYPE erlref SYSTEM "erlref.dtd"> + +<erlref> + <header> + <copyright> + <year>2017</year><year>2017</year> + <holder>Ericsson AB. All Rights Reserved.</holder> + </copyright> + <legalnotice> + Licensed under the Apache License, Version 2.0 (the "License"); + you may not use this file except in compliance with the License. + You may obtain a copy of the License at + + http://www.apache.org/licenses/LICENSE-2.0 + + Unless required by applicable law or agreed to in writing, software + distributed under the License is distributed on an "AS IS" BASIS, + WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. + See the License for the specific language governing permissions and + limitations under the License. + </legalnotice> + + <title>uri_string</title> + <prepared>Péter Dimitrov</prepared> + <docno>1</docno> + <date>2017-10-24</date> + <rev>A</rev> + </header> + <module>uri_string</module> + <modulesummary>URI processing functions.</modulesummary> + <description> + <p>This module contains functions for parsing and handling URIs + (<url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url>). + </p> + <p>A URI is an identifier consisting of a sequence of characters matching the syntax + rule named <em>URI</em> in <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url>. + </p> + <p> The generic URI syntax consists of a hierarchical sequence of components referred + to as the scheme, authority, path, query, and fragment:</p> + <pre> + URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ] + hier-part = "//" authority path-abempty + / path-absolute + / path-rootless + / path-empty + scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) + authority = [ userinfo "@" ] host [ ":" port ] + userinfo = *( unreserved / pct-encoded / sub-delims / ":" ) + + reserved = gen-delims / sub-delims + gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@" + sub-delims = "!" / "$" / "&" / "'" / "(" / ")" + / "*" / "+" / "," / ";" / "=" + + unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~" + </pre><br></br> + <p>The interpretation of a URI depends only on the characters used and not on how those + characters are represented in a network protocol.</p> + <p>The functions implemented by this module cover the following use cases:</p> + <list type="bulleted"> + <item>Parsing URIs into its components and returing a map<br></br> + <seealso marker="#parse/1"><c>parse/1</c></seealso> + </item> + <item>Recomposing a map of URI components into a URI string<br></br> + <seealso marker="#recompose/1"><c>recompose/1</c></seealso> + </item> + <item>Changing inbound binary and percent-encoding of URIs<br></br> + <seealso marker="#transcode/2"><c>transcode/2</c></seealso> + </item> + <item>Transforming URIs into a normalized form<br></br> + <seealso marker="#normalize/1"><c>normalize/1</c></seealso> + </item> + </list> + <p>There are four different encodings present during the handling of URIs:</p> + <list type="bulleted"> + <item>Inbound binary encoding in binaries</item> + <item>Inbound percent-encoding in lists and binaries</item> + <item>Outbound binary encoding in binaries</item> + <item>Outbound percent-encoding in lists and binaries</item> + </list> + <p>Functions with <c>uri_string()</c> argument accept lists, binaries and + mixed lists (lists with binary elements) as input type. All of the functions but + <c>transcode/2</c> expects input as lists of unicode codepoints, UTF-8 encoded binaries + and UTF-8 percent-encoded URI parts ("%C3%B6" corresponds to the unicode character "ö").</p> + <p>Unless otherwise specified the return value type and encoding are the same as the input + type and encoding. That is, binary input returns binary output, list input returns a list + output but mixed input returns list output.</p> + <p>In case of lists there is only percent-encoding. In binaries, however, both binary encoding + and percent-encoding shall be considered. <c>transcode/2</c> provides the means to convert + between the supported encodings, it takes a <c>uri_string()</c> and a list of options + specifying inbound and outbound encodings.</p> + <p><url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url> does not mandate any specific + character encoding and it is usually defined by the protocol or surrounding text. This library + takes the same assumption, binary and percent-encoding are handled as one configuration unit, + they cannot be set to different values.</p> + </description> + + <datatypes> + <datatype> + <name name="error"/> + <desc> + <p>Error tuple indicating the type of error. Possible values of the second component:</p> + <list type="bulleted"> + <item><c>invalid_input</c></item> + <item><c>invalid_map</c></item> + <item><c>invalid_percent_encoding</c></item> + <item><c>invalid_scheme</c></item> + <item><c>invalid_uri</c></item> + <item><c>invalid_utf8</c></item> + </list> + <p>The third component is a term providing additional information about the + cause of the error.</p> + </desc> + </datatype> + <datatype> + <name name="uri_map"/> + <desc> + <p>Map holding the main components of a URI.</p> + </desc> + </datatype> + <datatype> + <name name="uri_string"/> + <desc> + <p>List of unicode codepoints, a UTF-8 encoded binary, or a mix of the two, + representing an <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url> + compliant URI (<em>percent-encoded form</em>). + A URI is a sequence of characters from a very limited set: the letters of + the basic Latin alphabet, digits, and a few special characters.</p> + </desc> + </datatype> + </datatypes> + + <funcs> + + <func> + <name name="normalize" arity="1"/> + <fsummary>Syntax-based normalization.</fsummary> + <desc> + <p>Transforms <c><anno>URIString</anno></c> into a normalized form + using Syntax-Based Normalization as defined by + <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url>.</p> + <p>This function implements case normalization, percent-encoding + normalization, path segment normalization and scheme based normalization + for HTTP(S) with basic support for FTP, SSH, SFTP and TFTP.</p> + <p><em>Example:</em></p> + <pre> +1> <input>uri_string:normalize("/a/b/c/./../../g").</input> +"/a/g" +2> <![CDATA[uri_string:normalize(<<"mid/content=5/../6">>).]]> +<![CDATA[<<"mid/6">>]]> +3> uri_string:normalize("http://localhost:80"). +"https://localhost/" + </pre> + </desc> + </func> + + <func> + <name name="parse" arity="1"/> + <fsummary>Parse URI into a map.</fsummary> + <desc> + <p>Parses an <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url> + compliant <c>uri_string()</c> into a <c>uri_map()</c>, that holds the parsed + components of the <c>URI</c>. + If parsing fails, an error tuple is returned.</p> + <p>See also the opposite operation <seealso marker="#recompose/1"> + <c>recompose/1</c></seealso>.</p> + <p><em>Example:</em></p> + <pre> +1> <input>uri_string:parse("foo://[email protected]:8042/over/there?name=ferret#nose").</input> +#{fragment => "nose",host => "example.com", + path => "/over/there",port => 8042,query => "name=ferret", + scheme => foo,userinfo => "user"} +2> <![CDATA[uri_string:parse(<<"foo://[email protected]:8042/over/there?name=ferret">>).]]> +<![CDATA[#{host => <<"example.com">>,path => <<"/over/there">>, + port => 8042,query => <<"name=ferret">>,scheme => <<"foo">>, + userinfo => <<"user">>}]]> + </pre> + </desc> + </func> + + <func> + <name name="recompose" arity="1"/> + <fsummary>Recompose URI.</fsummary> + <desc> + <p>Creates an <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url> compliant + <c><anno>URIString</anno></c> (percent-encoded), based on the components of + <c><anno>URIMap</anno></c>. + If the <c><anno>URIMap</anno></c> is invalid, an error tuple is returned.</p> + <p>See also the opposite operation <seealso marker="#parse/1"> + <c>parse/1</c></seealso>.</p> + <p><em>Example:</em></p> + <pre> +1> <input>URIMap = #{fragment => "nose", host => "example.com", path => "/over/there",</input> +1> port => 8042, query => "name=ferret", scheme => "foo", userinfo => "user"}. +#{fragment => "top",host => "example.com", + path => "/over/there",port => 8042,query => "?name=ferret", + scheme => foo,userinfo => "user"} + +2> <input>uri_string:recompose(URIMap).</input> +"foo://example.com:8042/over/there?name=ferret#nose"</pre> + </desc> + </func> + + <func> + <name name="transcode" arity="2"/> + <fsummary>Transcode URI.</fsummary> + <desc> + <p>Transcodes an <url href="https://www.ietf.org/rfc/rfc3986.txt">RFC 3986</url> + compliant <c><anno>URIString</anno></c>, + where <c><anno>Options</anno></c> is a list of tagged tuples, specifying the inbound + (<c>in_encoding</c>) and outbound (<c>out_encoding</c>) encodings. <c>in_encoding</c> + and <c>out_encoding</c> specifies both binary encoding and percent-encoding for the + input and output data. Mixed encoding, where binary encoding is not the same as + percent-encoding, is not supported. + If an argument is invalid, an error tuple is returned.</p> + <p><em>Example:</em></p> + <pre> +1> <input><![CDATA[uri_string:transcode(<<"foo%00%00%00%F6bar"/utf32>>,]]></input> +1> [{in_encoding, utf32},{out_encoding, utf8}]). +<![CDATA[<<"foo%C3%B6bar"/utf8>>]]> +2> uri_string:transcode("foo%F6bar", [{in_encoding, latin1}, +2> {out_encoding, utf8}]). +"foo%C3%B6bar" + </pre> + </desc> + </func> + + </funcs> +</erlref> diff --git a/lib/stdlib/src/Makefile b/lib/stdlib/src/Makefile index bf836203ec..8b156929d7 100644 --- a/lib/stdlib/src/Makefile +++ b/lib/stdlib/src/Makefile @@ -121,6 +121,7 @@ MODULES= \ timer \ unicode \ unicode_util \ + uri_string \ win32reg \ zip diff --git a/lib/stdlib/src/stdlib.app.src b/lib/stdlib/src/stdlib.app.src index ab0824ca17..5fb48acfab 100644 --- a/lib/stdlib/src/stdlib.app.src +++ b/lib/stdlib/src/stdlib.app.src @@ -101,6 +101,7 @@ timer, unicode, unicode_util, + uri_string, win32reg, zip]}, {registered,[timer_server,rsh_starter,take_over_monitor,pool_master, diff --git a/lib/stdlib/src/uri_string.erl b/lib/stdlib/src/uri_string.erl new file mode 100644 index 0000000000..22212da222 --- /dev/null +++ b/lib/stdlib/src/uri_string.erl @@ -0,0 +1,1842 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +%% +%% [RFC 3986, Chapter 2.2. Reserved Characters] +%% +%% reserved = gen-delims / sub-delims +%% +%% gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@" +%% +%% sub-delims = "!" / "$" / "&" / "'" / "(" / ")" +%% / "*" / "+" / "," / ";" / "=" +%% +%% +%% [RFC 3986, Chapter 2.3. Unreserved Characters] +%% +%% unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~" +%% +%% +%% [RFC 3986, Chapter 3. Syntax Components] +%% +%% The generic URI syntax consists of a hierarchical sequence of +%% components referred to as the scheme, authority, path, query, and +%% fragment. +%% +%% URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ] +%% +%% hier-part = "//" authority path-abempty +%% / path-absolute +%% / path-rootless +%% / path-empty +%% +%% The scheme and path components are required, though the path may be +%% empty (no characters). When authority is present, the path must +%% either be empty or begin with a slash ("/") character. When +%% authority is not present, the path cannot begin with two slash +%% characters ("//"). These restrictions result in five different ABNF +%% rules for a path (Section 3.3), only one of which will match any +%% given URI reference. +%% +%% The following are two example URIs and their component parts: +%% +%% foo://example.com:8042/over/there?name=ferret#nose +%% \_/ \______________/\_________/ \_________/ \__/ +%% | | | | | +%% scheme authority path query fragment +%% | _____________________|__ +%% / \ / \ +%% urn:example:animal:ferret:nose +%% +%% +%% [RFC 3986, Chapter 3.1. Scheme] +%% +%% Each URI begins with a scheme name that refers to a specification for +%% assigning identifiers within that scheme. +%% +%% scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) +%% +%% +%% [RFC 3986, Chapter 3.2. Authority] +%% +%% Many URI schemes include a hierarchical element for a naming +%% authority so that governance of the name space defined by the +%% remainder of the URI is delegated to that authority (which may, in +%% turn, delegate it further). +%% +%% authority = [ userinfo "@" ] host [ ":" port ] +%% +%% +%% [RFC 3986, Chapter 3.2.1. User Information] +%% +%% The userinfo subcomponent may consist of a user name and, optionally, +%% scheme-specific information about how to gain authorization to access +%% the resource. The user information, if present, is followed by a +%% commercial at-sign ("@") that delimits it from the host. +%% +%% userinfo = *( unreserved / pct-encoded / sub-delims / ":" ) +%% +%% +%% [RFC 3986, Chapter 3.2.2. Host] +%% +%% The host subcomponent of authority is identified by an IP literal +%% encapsulated within square brackets, an IPv4 address in dotted- +%% decimal form, or a registered name. +%% +%% host = IP-literal / IPv4address / reg-name +%% +%% IP-literal = "[" ( IPv6address / IPvFuture ) "]" +%% +%% IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" ) +%% +%% IPv6address = 6( h16 ":" ) ls32 +%% / "::" 5( h16 ":" ) ls32 +%% / [ h16 ] "::" 4( h16 ":" ) ls32 +%% / [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32 +%% / [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32 +%% / [ *3( h16 ":" ) h16 ] "::" h16 ":" ls32 +%% / [ *4( h16 ":" ) h16 ] "::" ls32 +%% / [ *5( h16 ":" ) h16 ] "::" h16 +%% / [ *6( h16 ":" ) h16 ] "::" +%% +%% ls32 = ( h16 ":" h16 ) / IPv4address +%% ; least-significant 32 bits of address +%% +%% h16 = 1*4HEXDIG +%% ; 16 bits of address represented in hexadecimal +%% +%% IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet +%% +%% dec-octet = DIGIT ; 0-9 +%% / %x31-39 DIGIT ; 10-99 +%% / "1" 2DIGIT ; 100-199 +%% / "2" %x30-34 DIGIT ; 200-249 +%% / "25" %x30-35 ; 250-255 +%% +%% reg-name = *( unreserved / pct-encoded / sub-delims ) +%% +%% +%% [RFC 3986, Chapter 3.2.2. Port] +%% +%% The port subcomponent of authority is designated by an optional port +%% number in decimal following the host and delimited from it by a +%% single colon (":") character. +%% +%% port = *DIGIT +%% +%% +%% [RFC 3986, Chapter 3.3. Path] +%% +%% The path component contains data, usually organized in hierarchical +%% form, that, along with data in the non-hierarchical query component +%% (Section 3.4), serves to identify a resource within the scope of the +%% URI's scheme and naming authority (if any). The path is terminated +%% by the first question mark ("?") or number sign ("#") character, or +%% by the end of the URI. +%% +%% path = path-abempty ; begins with "/" or is empty +%% / path-absolute ; begins with "/" but not "//" +%% / path-noscheme ; begins with a non-colon segment +%% / path-rootless ; begins with a segment +%% / path-empty ; zero characters +%% +%% path-abempty = *( "/" segment ) +%% path-absolute = "/" [ segment-nz *( "/" segment ) ] +%% path-noscheme = segment-nz-nc *( "/" segment ) +%% path-rootless = segment-nz *( "/" segment ) +%% path-empty = 0<pchar> +%% segment = *pchar +%% segment-nz = 1*pchar +%% segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" ) +%% ; non-zero-length segment without any colon ":" +%% +%% pchar = unreserved / pct-encoded / sub-delims / ":" / "@" +%% +%% +%% [RFC 3986, Chapter 3.4. Query] +%% +%% The query component contains non-hierarchical data that, along with +%% data in the path component (Section 3.3), serves to identify a +%% resource within the scope of the URI's scheme and naming authority +%% (if any). The query component is indicated by the first question +%% mark ("?") character and terminated by a number sign ("#") character +%% or by the end of the URI. +%% +%% query = *( pchar / "/" / "?" ) +%% +%% +%% [RFC 3986, Chapter 3.5. Fragment] +%% +%% The fragment identifier component of a URI allows indirect +%% identification of a secondary resource by reference to a primary +%% resource and additional identifying information. +%% +%% fragment = *( pchar / "/" / "?" ) +%% +%% +%% [RFC 3986, Chapter 4.1. URI Reference] +%% +%% URI-reference is used to denote the most common usage of a resource +%% identifier. +%% +%% URI-reference = URI / relative-ref +%% +%% +%% [RFC 3986, Chapter 4.2. Relative Reference] +%% +%% A relative reference takes advantage of the hierarchical syntax +%% (Section 1.2.3) to express a URI reference relative to the name space +%% of another hierarchical URI. +%% +%% relative-ref = relative-part [ "?" query ] [ "#" fragment ] +%% +%% relative-part = "//" authority path-abempty +%% / path-absolute +%% / path-noscheme +%% / path-empty +%% +%% +%% [RFC 3986, Chapter 4.3. Absolute URI] +%% +%% Some protocol elements allow only the absolute form of a URI without +%% a fragment identifier. For example, defining a base URI for later +%% use by relative references calls for an absolute-URI syntax rule that +%% does not allow a fragment. +%% +%% absolute-URI = scheme ":" hier-part [ "?" query ] +%% +-module(uri_string). + +%%------------------------------------------------------------------------- +%% External API +%%------------------------------------------------------------------------- +-export([normalize/1, parse/1, + recompose/1, transcode/2]). +-export_type([error/0, uri_map/0, uri_string/0]). + + +%%------------------------------------------------------------------------- +%% Internal API +%%------------------------------------------------------------------------- +-export([is_host/1, is_path/1]). % suppress warnings + + +%%------------------------------------------------------------------------- +%% Macros +%%------------------------------------------------------------------------- +-define(CHAR(Char), <<Char/utf8>>). +-define(STRING_EMPTY, <<>>). +-define(STRING(MatchStr), <<MatchStr/binary>>). +-define(STRING_REST(MatchStr, Rest), <<MatchStr/utf8, Rest/binary>>). + +-define(DEC2HEX(X), + if ((X) >= 0) andalso ((X) =< 9) -> (X) + $0; + ((X) >= 10) andalso ((X) =< 15) -> (X) + $A - 10 + end). + +-define(HEX2DEC(X), + if ((X) >= $0) andalso ((X) =< $9) -> (X) - $0; + ((X) >= $A) andalso ((X) =< $F) -> (X) - $A + 10; + ((X) >= $a) andalso ((X) =< $f) -> (X) - $a + 10 + end). + + +%%%========================================================================= +%%% API +%%%========================================================================= + +%%------------------------------------------------------------------------- +%% URI compliant with RFC 3986 +%% ASCII %x21 - %x7A ("!" - "z") except +%% %x34 " double quote +%% %x60 < less than +%% %x62 > greater than +%% %x92 \ backslash +%% %x94 ^ caret / circumflex +%% %x96 ` grave / accent +%%------------------------------------------------------------------------- +-type uri_string() :: iodata(). +-type error() :: {error, atom(), term()}. + + +%%------------------------------------------------------------------------- +%% RFC 3986, Chapter 3. Syntax Components +%%------------------------------------------------------------------------- +-type uri_map() :: + #{fragment => unicode:chardata(), + host => unicode:chardata(), + path => unicode:chardata(), + port => non_neg_integer() | undefined, + query => unicode:chardata(), + scheme => unicode:chardata(), + userinfo => unicode:chardata()} | #{}. + + +%%------------------------------------------------------------------------- +%% Normalize URIs +%%------------------------------------------------------------------------- +-spec normalize(URIString) -> NormalizedURI when + URIString :: uri_string(), + NormalizedURI :: uri_string(). +normalize(URIString) -> + %% Percent-encoding normalization and case normalization for + %% percent-encoded triplets are achieved by running parse and + %% recompose on the input URI string. + recompose( + normalize_path_segment( + normalize_scheme_based( + normalize_case( + parse(URIString))))). + + +%%------------------------------------------------------------------------- +%% Parse URIs +%%------------------------------------------------------------------------- +-spec parse(URIString) -> URIMap when + URIString :: uri_string(), + URIMap :: uri_map() + | error(). +parse(URIString) when is_binary(URIString) -> + try parse_uri_reference(URIString, #{}) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end; +parse(URIString) when is_list(URIString) -> + try + Binary = unicode:characters_to_binary(URIString), + Map = parse_uri_reference(Binary, #{}), + convert_mapfields_to_list(Map) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end. + + +%%------------------------------------------------------------------------- +%% Recompose URIs +%%------------------------------------------------------------------------- +-spec recompose(URIMap) -> URIString when + URIMap :: uri_map(), + URIString :: uri_string() + | error(). +recompose(Map) -> + case is_valid_map(Map) of + false -> + {error, invalid_map, Map}; + true -> + try + T0 = update_scheme(Map, empty), + T1 = update_userinfo(Map, T0), + T2 = update_host(Map, T1), + T3 = update_port(Map, T2), + T4 = update_path(Map, T3), + T5 = update_query(Map, T4), + update_fragment(Map, T5) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end + end. + + +%%------------------------------------------------------------------------- +%% Transcode URIs +%%------------------------------------------------------------------------- +-spec transcode(URIString, Options) -> Result when + URIString :: uri_string(), + Options :: [{in_encoding, unicode:encoding()}|{out_encoding, unicode:encoding()}], + Result :: uri_string() + | error(). +transcode(URIString, Options) when is_binary(URIString) -> + try + InEnc = proplists:get_value(in_encoding, Options, utf8), + OutEnc = proplists:get_value(out_encoding, Options, utf8), + List = convert_to_list(URIString, InEnc), + Output = transcode(List, [], InEnc, OutEnc), + convert_to_binary(Output, utf8, OutEnc) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end; +transcode(URIString, Options) when is_list(URIString) -> + InEnc = proplists:get_value(in_encoding, Options, utf8), + OutEnc = proplists:get_value(out_encoding, Options, utf8), + Flattened = flatten_list(URIString, InEnc), + try transcode(Flattened, [], InEnc, OutEnc) + catch + throw:{error, Atom, RestData} -> {error, Atom, RestData} + end. + + +%%%======================================================================== +%%% Internal functions +%%%======================================================================== + +%%------------------------------------------------------------------------- +%% Converts Map fields to lists +%%------------------------------------------------------------------------- +convert_mapfields_to_list(Map) -> + Fun = fun (_, V) when is_binary(V) -> unicode:characters_to_list(V); + (_, V) -> V end, + maps:map(Fun, Map). + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 4.1. URI Reference] +%% +%% URI-reference is used to denote the most common usage of a resource +%% identifier. +%% +%% URI-reference = URI / relative-ref +%%------------------------------------------------------------------------- +-spec parse_uri_reference(binary(), uri_map()) -> uri_map(). +parse_uri_reference(<<>>, _) -> #{path => <<>>}; +parse_uri_reference(URIString, URI) -> + try parse_scheme_start(URIString, URI) + catch + throw:{_,_,_} -> + parse_relative_part(URIString, URI) + end. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 4.2. Relative Reference] +%% +%% A relative reference takes advantage of the hierarchical syntax +%% (Section 1.2.3) to express a URI reference relative to the name space +%% of another hierarchical URI. +%% +%% relative-ref = relative-part [ "?" query ] [ "#" fragment ] +%% +%% relative-part = "//" authority path-abempty +%% / path-absolute +%% / path-noscheme +%% / path-empty +%%------------------------------------------------------------------------- +-spec parse_relative_part(binary(), uri_map()) -> uri_map(). +parse_relative_part(?STRING_REST("//", Rest), URI) -> + %% Parse userinfo - "//" is NOT part of authority + try parse_userinfo(Rest, URI) of + {T, URI1} -> + Userinfo = calculate_parsed_userinfo(Rest, T), + URI2 = maybe_add_path(URI1), + URI2#{userinfo => decode_userinfo(Userinfo)} + catch + throw:{_,_,_} -> + {T, URI1} = parse_host(Rest, URI), + Host = calculate_parsed_host_port(Rest, T), + URI2 = maybe_add_path(URI1), + URI2#{host => decode_host(remove_brackets(Host))} + end; +parse_relative_part(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-absolute + Path = calculate_parsed_part(Rest, T), + URI1#{path => decode_path(?STRING_REST($/, Path))}; +parse_relative_part(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + URI2 = maybe_add_path(URI1), + URI2#{query => decode_query(Query)}; +parse_relative_part(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + URI2 = maybe_add_path(URI1), + URI2#{fragment => decode_fragment(Fragment)}; +parse_relative_part(?STRING_REST(Char, Rest), URI) -> + case is_segment_nz_nc(Char) of + true -> + {T, URI1} = parse_segment_nz_nc(Rest, URI), % path-noscheme + Path = calculate_parsed_part(Rest, T), + URI1#{path => decode_path(?STRING_REST(Char, Path))}; + false -> throw({error,invalid_uri,[Char]}) + end. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.3. Path] +%% +%% The path component contains data, usually organized in hierarchical +%% form, that, along with data in the non-hierarchical query component +%% (Section 3.4), serves to identify a resource within the scope of the +%% URI's scheme and naming authority (if any). The path is terminated +%% by the first question mark ("?") or number sign ("#") character, or +%% by the end of the URI. +%% +%% path = path-abempty ; begins with "/" or is empty +%% / path-absolute ; begins with "/" but not "//" +%% / path-noscheme ; begins with a non-colon segment +%% / path-rootless ; begins with a segment +%% / path-empty ; zero characters +%% +%% path-abempty = *( "/" segment ) +%% path-absolute = "/" [ segment-nz *( "/" segment ) ] +%% path-noscheme = segment-nz-nc *( "/" segment ) +%% path-rootless = segment-nz *( "/" segment ) +%% path-empty = 0<pchar> +%% segment = *pchar +%% segment-nz = 1*pchar +%% segment-nz-nc = 1*( unreserved / pct-encoded / sub-delims / "@" ) +%% ; non-zero-length segment without any colon ":" +%% +%% pchar = unreserved / pct-encoded / sub-delims / ":" / "@" +%%------------------------------------------------------------------------- + +%%------------------------------------------------------------------------- +%% path-abempty +%%------------------------------------------------------------------------- +-spec parse_segment(binary(), uri_map()) -> {binary(), uri_map()}. +parse_segment(?STRING_REST($/, Rest), URI) -> + parse_segment(Rest, URI); % segment +parse_segment(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_segment(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_segment(?STRING_REST(Char, Rest), URI) -> + case is_pchar(Char) of + true -> parse_segment(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_segment(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +%%------------------------------------------------------------------------- +%% path-noscheme +%%------------------------------------------------------------------------- +-spec parse_segment_nz_nc(binary(), uri_map()) -> {binary(), uri_map()}. +parse_segment_nz_nc(?STRING_REST($/, Rest), URI) -> + parse_segment(Rest, URI); % segment +parse_segment_nz_nc(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_segment_nz_nc(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_segment_nz_nc(?STRING_REST(Char, Rest), URI) -> + case is_segment_nz_nc(Char) of + true -> parse_segment_nz_nc(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_segment_nz_nc(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +%% Check if char is pchar. +-spec is_pchar(char()) -> boolean(). +is_pchar($%) -> true; % pct-encoded +is_pchar($:) -> true; +is_pchar($@) -> true; +is_pchar(Char) -> is_unreserved(Char) orelse is_sub_delim(Char). + +%% Check if char is segment_nz_nc. +-spec is_segment_nz_nc(char()) -> boolean(). +is_segment_nz_nc($%) -> true; % pct-encoded +is_segment_nz_nc($@) -> true; +is_segment_nz_nc(Char) -> is_unreserved(Char) orelse is_sub_delim(Char). + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.1. Scheme] +%% +%% Each URI begins with a scheme name that refers to a specification for +%% assigning identifiers within that scheme. +%% +%% scheme = ALPHA *( ALPHA / DIGIT / "+" / "-" / "." ) +%%------------------------------------------------------------------------- +-spec parse_scheme_start(binary(), uri_map()) -> uri_map(). +parse_scheme_start(?STRING_REST(Char, Rest), URI) -> + case is_alpha(Char) of + true -> {T, URI1} = parse_scheme(Rest, URI), + Scheme = calculate_parsed_scheme(Rest, T), + URI2 = maybe_add_path(URI1), + URI2#{scheme => ?STRING_REST(Char, Scheme)}; + false -> throw({error,invalid_uri,[Char]}) + end. + +%% Add path component if it missing after parsing the URI. +%% According to the URI specification there is always a +%% path component in every URI-reference and it can be +%% empty. +maybe_add_path(Map) -> + case maps:is_key(path, Map) of + false -> + Map#{path => <<>>}; + _Else -> + Map + end. + + +-spec parse_scheme(binary(), uri_map()) -> {binary(), uri_map()}. +parse_scheme(?STRING_REST($:, Rest), URI) -> + {_, URI1} = parse_hier(Rest, URI), + {Rest, URI1}; +parse_scheme(?STRING_REST(Char, Rest), URI) -> + case is_scheme(Char) of + true -> parse_scheme(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_scheme(?STRING_EMPTY, _URI) -> + throw({error,invalid_uri,<<>>}). + + +%% Check if char is allowed in scheme +-spec is_scheme(char()) -> boolean(). +is_scheme($+) -> true; +is_scheme($-) -> true; +is_scheme($.) -> true; +is_scheme(Char) -> is_alpha(Char) orelse is_digit(Char). + + +%%------------------------------------------------------------------------- +%% hier-part = "//" authority path-abempty +%% / path-absolute +%% / path-rootless +%% / path-empty +%%------------------------------------------------------------------------- +-spec parse_hier(binary(), uri_map()) -> {binary(), uri_map()}. +parse_hier(?STRING_REST("//", Rest), URI) -> + % Parse userinfo - "//" is NOT part of authority + try parse_userinfo(Rest, URI) of + {T, URI1} -> + Userinfo = calculate_parsed_userinfo(Rest, T), + {Rest, URI1#{userinfo => decode_userinfo(Userinfo)}} + catch + throw:{_,_,_} -> + {T, URI1} = parse_host(Rest, URI), + Host = calculate_parsed_host_port(Rest, T), + {Rest, URI1#{host => decode_host(remove_brackets(Host))}} + end; +parse_hier(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-absolute + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_hier(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_hier(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_hier(?STRING_REST(Char, Rest), URI) -> % path-rootless + case is_pchar(Char) of + true -> % segment_nz + {T, URI1} = parse_segment(Rest, URI), + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST(Char, Path))}}; + false -> throw({error,invalid_uri,[Char]}) + end; +parse_hier(?STRING_EMPTY, URI) -> + {<<>>, URI}. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.2. Authority] +%% +%% Many URI schemes include a hierarchical element for a naming +%% authority so that governance of the name space defined by the +%% remainder of the URI is delegated to that authority (which may, in +%% turn, delegate it further). +%% +%% The authority component is preceded by a double slash ("//") and is +%% terminated by the next slash ("/"), question mark ("?"), or number +%% sign ("#") character, or by the end of the URI. +%% +%% authority = [ userinfo "@" ] host [ ":" port ] +%% +%% +%% [RFC 3986, Chapter 3.2.1. User Information] +%% +%% The userinfo subcomponent may consist of a user name and, optionally, +%% scheme-specific information about how to gain authorization to access +%% the resource. The user information, if present, is followed by a +%% commercial at-sign ("@") that delimits it from the host. +%% +%% userinfo = *( unreserved / pct-encoded / sub-delims / ":" ) +%%------------------------------------------------------------------------- +-spec parse_userinfo(binary(), uri_map()) -> {binary(), uri_map()}. +parse_userinfo(?CHAR($@), URI) -> + {?STRING_EMPTY, URI#{host => <<>>}}; +parse_userinfo(?STRING_REST($@, Rest), URI) -> + {T, URI1} = parse_host(Rest, URI), + Host = calculate_parsed_host_port(Rest, T), + {Rest, URI1#{host => decode_host(remove_brackets(Host))}}; +parse_userinfo(?STRING_REST(Char, Rest), URI) -> + case is_userinfo(Char) of + true -> parse_userinfo(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_userinfo(?STRING_EMPTY, _URI) -> + %% URI cannot end in userinfo state + throw({error,invalid_uri,<<>>}). + + +%% Check if char is allowed in userinfo +-spec is_userinfo(char()) -> boolean(). +is_userinfo($%) -> true; % pct-encoded +is_userinfo($:) -> true; +is_userinfo(Char) -> is_unreserved(Char) orelse is_sub_delim(Char). + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.2.2. Host] +%% +%% The host subcomponent of authority is identified by an IP literal +%% encapsulated within square brackets, an IPv4 address in dotted- +%% decimal form, or a registered name. +%% +%% host = IP-literal / IPv4address / reg-name +%% +%% IP-literal = "[" ( IPv6address / IPvFuture ) "]" +%% +%% IPvFuture = "v" 1*HEXDIG "." 1*( unreserved / sub-delims / ":" ) +%% +%% IPv6address = 6( h16 ":" ) ls32 +%% / "::" 5( h16 ":" ) ls32 +%% / [ h16 ] "::" 4( h16 ":" ) ls32 +%% / [ *1( h16 ":" ) h16 ] "::" 3( h16 ":" ) ls32 +%% / [ *2( h16 ":" ) h16 ] "::" 2( h16 ":" ) ls32 +%% / [ *3( h16 ":" ) h16 ] "::" h16 ":" ls32 +%% / [ *4( h16 ":" ) h16 ] "::" ls32 +%% / [ *5( h16 ":" ) h16 ] "::" h16 +%% / [ *6( h16 ":" ) h16 ] "::" +%% +%% ls32 = ( h16 ":" h16 ) / IPv4address +%% ; least-significant 32 bits of address +%% +%% h16 = 1*4HEXDIG +%% ; 16 bits of address represented in hexadecimal +%% +%% IPv4address = dec-octet "." dec-octet "." dec-octet "." dec-octet +%% +%% dec-octet = DIGIT ; 0-9 +%% / %x31-39 DIGIT ; 10-99 +%% / "1" 2DIGIT ; 100-199 +%% / "2" %x30-34 DIGIT ; 200-249 +%% / "25" %x30-35 ; 250-255 +%% +%% reg-name = *( unreserved / pct-encoded / sub-delims ) +%%------------------------------------------------------------------------- +-spec parse_host(binary(), uri_map()) -> {binary(), uri_map()}. +parse_host(?STRING_REST($:, Rest), URI) -> + {T, URI1} = parse_port(Rest, URI), + H = calculate_parsed_host_port(Rest, T), + Port = get_port(H), + {Rest, URI1#{port => Port}}; +parse_host(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-abempty + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_host(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_host(?STRING_REST($[, Rest), URI) -> + parse_ipv6_bin(Rest, [], URI); +parse_host(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_host(?STRING_REST(Char, Rest), URI) -> + case is_digit(Char) of + true -> parse_ipv4_bin(Rest, [Char], URI); + false -> parse_reg_name(?STRING_REST(Char, Rest), URI) + end; +parse_host(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +-spec parse_reg_name(binary(), uri_map()) -> {binary(), uri_map()}. +parse_reg_name(?STRING_REST($:, Rest), URI) -> + {T, URI1} = parse_port(Rest, URI), + H = calculate_parsed_host_port(Rest, T), + Port = get_port(H), + {Rest, URI1#{port => Port}}; +parse_reg_name(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-abempty + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_reg_name(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_reg_name(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_reg_name(?STRING_REST(Char, Rest), URI) -> + case is_reg_name(Char) of + true -> parse_reg_name(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_reg_name(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + +%% Check if char is allowed in reg-name +-spec is_reg_name(char()) -> boolean(). +is_reg_name($%) -> true; +is_reg_name(Char) -> is_unreserved(Char) orelse is_sub_delim(Char). + + +-spec parse_ipv4_bin(binary(), list(), uri_map()) -> {binary(), uri_map()}. +parse_ipv4_bin(?STRING_REST($:, Rest), Acc, URI) -> + _ = validate_ipv4_address(lists:reverse(Acc)), + {T, URI1} = parse_port(Rest, URI), + H = calculate_parsed_host_port(Rest, T), + Port = get_port(H), + {Rest, URI1#{port => Port}}; +parse_ipv4_bin(?STRING_REST($/, Rest), Acc, URI) -> + _ = validate_ipv4_address(lists:reverse(Acc)), + {T, URI1} = parse_segment(Rest, URI), % path-abempty + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_ipv4_bin(?STRING_REST($?, Rest), Acc, URI) -> + _ = validate_ipv4_address(lists:reverse(Acc)), + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_ipv4_bin(?STRING_REST($#, Rest), Acc, URI) -> + _ = validate_ipv4_address(lists:reverse(Acc)), + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_ipv4_bin(?STRING_REST(Char, Rest), Acc, URI) -> + case is_ipv4(Char) of + true -> parse_ipv4_bin(Rest, [Char|Acc], URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_ipv4_bin(?STRING_EMPTY, Acc, URI) -> + _ = validate_ipv4_address(lists:reverse(Acc)), + {?STRING_EMPTY, URI}. + + +%% Check if char is allowed in IPv4 addresses +-spec is_ipv4(char()) -> boolean(). +is_ipv4($.) -> true; +is_ipv4(Char) -> is_digit(Char). + +-spec validate_ipv4_address(list()) -> list(). +validate_ipv4_address(Addr) -> + case inet:parse_ipv4strict_address(Addr) of + {ok, _} -> Addr; + {error, _} -> throw({error,invalid_uri,Addr}) + end. + + +-spec parse_ipv6_bin(binary(), list(), uri_map()) -> {binary(), uri_map()}. +parse_ipv6_bin(?STRING_REST($], Rest), Acc, URI) -> + _ = validate_ipv6_address(lists:reverse(Acc)), + parse_ipv6_bin_end(Rest, URI); +parse_ipv6_bin(?STRING_REST(Char, Rest), Acc, URI) -> + case is_ipv6(Char) of + true -> parse_ipv6_bin(Rest, [Char|Acc], URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_ipv6_bin(?STRING_EMPTY, _Acc, _URI) -> + throw({error,invalid_uri,<<>>}). + +%% Check if char is allowed in IPv6 addresses +-spec is_ipv6(char()) -> boolean(). +is_ipv6($:) -> true; +is_ipv6($.) -> true; +is_ipv6(Char) -> is_hex_digit(Char). + + +-spec parse_ipv6_bin_end(binary(), uri_map()) -> {binary(), uri_map()}. +parse_ipv6_bin_end(?STRING_REST($:, Rest), URI) -> + {T, URI1} = parse_port(Rest, URI), + H = calculate_parsed_host_port(Rest, T), + Port = get_port(H), + {Rest, URI1#{port => Port}}; +parse_ipv6_bin_end(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-abempty + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_ipv6_bin_end(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_ipv6_bin_end(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_ipv6_bin_end(?STRING_REST(Char, Rest), URI) -> + case is_ipv6(Char) of + true -> parse_ipv6_bin_end(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_ipv6_bin_end(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + +-spec validate_ipv6_address(list()) -> list(). +validate_ipv6_address(Addr) -> + case inet:parse_ipv6strict_address(Addr) of + {ok, _} -> Addr; + {error, _} -> throw({error,invalid_uri,Addr}) + end. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.2.2. Port] +%% +%% The port subcomponent of authority is designated by an optional port +%% number in decimal following the host and delimited from it by a +%% single colon (":") character. +%% +%% port = *DIGIT +%%------------------------------------------------------------------------- +-spec parse_port(binary(), uri_map()) -> {binary(), uri_map()}. +parse_port(?STRING_REST($/, Rest), URI) -> + {T, URI1} = parse_segment(Rest, URI), % path-abempty + Path = calculate_parsed_part(Rest, T), + {Rest, URI1#{path => decode_path(?STRING_REST($/, Path))}}; +parse_port(?STRING_REST($?, Rest), URI) -> + {T, URI1} = parse_query(Rest, URI), % path-empty ?query + Query = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{query => decode_query(Query)}}; +parse_port(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), % path-empty + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_port(?STRING_REST(Char, Rest), URI) -> + case is_digit(Char) of + true -> parse_port(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_port(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.4. Query] +%% +%% The query component contains non-hierarchical data that, along with +%% data in the path component (Section 3.3), serves to identify a +%% resource within the scope of the URI's scheme and naming authority +%% (if any). The query component is indicated by the first question +%% mark ("?") character and terminated by a number sign ("#") character +%% or by the end of the URI. +%% +%% query = *( pchar / "/" / "?" ) +%%------------------------------------------------------------------------- +-spec parse_query(binary(), uri_map()) -> {binary(), uri_map()}. +parse_query(?STRING_REST($#, Rest), URI) -> + {T, URI1} = parse_fragment(Rest, URI), + Fragment = calculate_parsed_query_fragment(Rest, T), + {Rest, URI1#{fragment => decode_fragment(Fragment)}}; +parse_query(?STRING_REST(Char, Rest), URI) -> + case is_query(Char) of + true -> parse_query(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_query(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +%% Check if char is allowed in query +-spec is_query(char()) -> boolean(). +is_query($/) -> true; +is_query($?) -> true; +is_query(Char) -> is_pchar(Char). + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 3.5. Fragment] +%% +%% The fragment identifier component of a URI allows indirect +%% identification of a secondary resource by reference to a primary +%% resource and additional identifying information. +%% +%% fragment = *( pchar / "/" / "?" ) +%%------------------------------------------------------------------------- +-spec parse_fragment(binary(), uri_map()) -> {binary(), uri_map()}. +parse_fragment(?STRING_REST(Char, Rest), URI) -> + case is_fragment(Char) of + true -> parse_fragment(Rest, URI); + false -> throw({error,invalid_uri,[Char]}) + end; +parse_fragment(?STRING_EMPTY, URI) -> + {?STRING_EMPTY, URI}. + + +%% Check if char is allowed in fragment +-spec is_fragment(char()) -> boolean(). +is_fragment($/) -> true; +is_fragment($?) -> true; +is_fragment(Char) -> is_pchar(Char). + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 2.2. Reserved Characters] +%% +%% reserved = gen-delims / sub-delims +%% +%% gen-delims = ":" / "/" / "?" / "#" / "[" / "]" / "@" +%% +%% sub-delims = "!" / "$" / "&" / "'" / "(" / ")" +%% / "*" / "+" / "," / ";" / "=" +%% +%%------------------------------------------------------------------------- + +%% Check if char is sub-delim. +-spec is_sub_delim(char()) -> boolean(). +is_sub_delim($!) -> true; +is_sub_delim($$) -> true; +is_sub_delim($&) -> true; +is_sub_delim($') -> true; +is_sub_delim($() -> true; +is_sub_delim($)) -> true; + +is_sub_delim($*) -> true; +is_sub_delim($+) -> true; +is_sub_delim($,) -> true; +is_sub_delim($;) -> true; +is_sub_delim($=) -> true; +is_sub_delim(_) -> false. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 2.3. Unreserved Characters] +%% +%% unreserved = ALPHA / DIGIT / "-" / "." / "_" / "~" +%% +%%------------------------------------------------------------------------- +-spec is_unreserved(char()) -> boolean(). +is_unreserved($-) -> true; +is_unreserved($.) -> true; +is_unreserved($_) -> true; +is_unreserved($~) -> true; +is_unreserved(Char) -> is_alpha(Char) orelse is_digit(Char). + +-spec is_alpha(char()) -> boolean(). +is_alpha(C) + when $A =< C, C =< $Z; + $a =< C, C =< $z -> true; +is_alpha(_) -> false. + +-spec is_digit(char()) -> boolean(). +is_digit(C) + when $0 =< C, C =< $9 -> true; +is_digit(_) -> false. + +-spec is_hex_digit(char()) -> boolean(). +is_hex_digit(C) + when $0 =< C, C =< $9;$a =< C, C =< $f;$A =< C, C =< $F -> true; +is_hex_digit(_) -> false. + + +%% Remove enclosing brackets from binary +-spec remove_brackets(binary()) -> binary(). +remove_brackets(<<$[/utf8, Rest/binary>>) -> + {H,T} = split_binary(Rest, byte_size(Rest) - 1), + case T =:= <<$]/utf8>> of + true -> H; + false -> Rest + end; +remove_brackets(Addr) -> Addr. + + +%%------------------------------------------------------------------------- +%% Helper functions for calculating the parsed binary. +%%------------------------------------------------------------------------- +-spec calculate_parsed_scheme(binary(), binary()) -> binary(). +calculate_parsed_scheme(Input, <<>>) -> + strip_last_char(Input, [$:]); +calculate_parsed_scheme(Input, Unparsed) -> + get_parsed_binary(Input, Unparsed). + + +-spec calculate_parsed_part(binary(), binary()) -> binary(). +calculate_parsed_part(Input, <<>>) -> + strip_last_char(Input, [$?,$#]); +calculate_parsed_part(Input, Unparsed) -> + get_parsed_binary(Input, Unparsed). + + +-spec calculate_parsed_userinfo(binary(), binary()) -> binary(). +calculate_parsed_userinfo(Input, <<>>) -> + strip_last_char(Input, [$?,$#,$@]); +calculate_parsed_userinfo(Input, Unparsed) -> + get_parsed_binary(Input, Unparsed). + + +-spec calculate_parsed_host_port(binary(), binary()) -> binary(). +calculate_parsed_host_port(Input, <<>>) -> + strip_last_char(Input, [$:,$?,$#,$/]); +calculate_parsed_host_port(Input, Unparsed) -> + get_parsed_binary(Input, Unparsed). + + +calculate_parsed_query_fragment(Input, <<>>) -> + strip_last_char(Input, [$#]); +calculate_parsed_query_fragment(Input, Unparsed) -> + get_parsed_binary(Input, Unparsed). + + +get_port(<<>>) -> + undefined; +get_port(B) -> + try binary_to_integer(B) + catch + error:badarg -> + throw({error, invalid_uri, B}) + end. + + +%% Strip last char if it is in list +%% +%% This function is optimized for speed: parse/1 is about 10% faster than +%% with an alternative implementation based on lists and sets. +strip_last_char(<<>>, _) -> <<>>; +strip_last_char(Input, [C0]) -> + case binary:last(Input) of + C0 -> + init_binary(Input); + _Else -> + Input + end; +strip_last_char(Input, [C0,C1]) -> + case binary:last(Input) of + C0 -> + init_binary(Input); + C1 -> + init_binary(Input); + _Else -> + Input + end; +strip_last_char(Input, [C0,C1,C2]) -> + case binary:last(Input) of + C0 -> + init_binary(Input); + C1 -> + init_binary(Input); + C2 -> + init_binary(Input); + _Else -> + Input + end; +strip_last_char(Input, [C0,C1,C2,C3]) -> + case binary:last(Input) of + C0 -> + init_binary(Input); + C1 -> + init_binary(Input); + C2 -> + init_binary(Input); + C3 -> + init_binary(Input); + _Else -> + Input + end. + + +%% Get parsed binary +get_parsed_binary(Input, Unparsed) -> + {First, _} = split_binary(Input, byte_size(Input) - byte_size_exl_head(Unparsed)), + First. + + +%% Return all bytes of the binary except the last one. The binary must be non-empty. +init_binary(B) -> + {Init, _} = + split_binary(B, byte_size(B) - 1), + Init. + + +%% Returns the size of a binary exluding the first element. +%% Used in calls to split_binary(). +-spec byte_size_exl_head(binary()) -> number(). +byte_size_exl_head(<<>>) -> 0; +byte_size_exl_head(Binary) -> byte_size(Binary) + 1. + + +%%------------------------------------------------------------------------- +%% [RFC 3986, Chapter 2.1. Percent-Encoding] +%% +%% A percent-encoding mechanism is used to represent a data octet in a +%% component when that octet's corresponding character is outside the +%% allowed set or is being used as a delimiter of, or within, the +%% component. A percent-encoded octet is encoded as a character +%% triplet, consisting of the percent character "%" followed by the two +%% hexadecimal digits representing that octet's numeric value. For +%% example, "%20" is the percent-encoding for the binary octet +%% "00100000" (ABNF: %x20), which in US-ASCII corresponds to the space +%% character (SP). Section 2.4 describes when percent-encoding and +%% decoding is applied. +%% +%% pct-encoded = "%" HEXDIG HEXDIG +%%------------------------------------------------------------------------- +-spec decode_userinfo(binary()) -> binary(). +decode_userinfo(Cs) -> + check_utf8(decode(Cs, fun is_userinfo/1, <<>>)). + +-spec decode_host(binary()) -> binary(). +decode_host(Cs) -> + check_utf8(decode(Cs, fun is_host/1, <<>>)). + +-spec decode_path(binary()) -> binary(). +decode_path(Cs) -> + check_utf8(decode(Cs, fun is_path/1, <<>>)). + +-spec decode_query(binary()) -> binary(). +decode_query(Cs) -> + check_utf8(decode(Cs, fun is_query/1, <<>>)). + +-spec decode_fragment(binary()) -> binary(). +decode_fragment(Cs) -> + check_utf8(decode(Cs, fun is_fragment/1, <<>>)). + + +%% Returns Cs if it is utf8 encoded. +check_utf8(Cs) -> + case unicode:characters_to_list(Cs) of + {incomplete,_,_} -> + throw({error,invalid_utf8,Cs}); + {error,_,_} -> + throw({error,invalid_utf8,Cs}); + _ -> Cs + end. + +%%------------------------------------------------------------------------- +%% Percent-encode +%%------------------------------------------------------------------------- + +%% Only validates as scheme cannot have percent-encoded characters +-spec encode_scheme(list()|binary()) -> list() | binary(). +encode_scheme([]) -> + throw({error,invalid_scheme,""}); +encode_scheme(<<>>) -> + throw({error,invalid_scheme,<<>>}); +encode_scheme(Scheme) -> + case validate_scheme(Scheme) of + true -> Scheme; + false -> throw({error,invalid_scheme,Scheme}) + end. + +-spec encode_userinfo(list()|binary()) -> list() | binary(). +encode_userinfo(Cs) -> + encode(Cs, fun is_userinfo/1). + +-spec encode_host(list()|binary()) -> list() | binary(). +encode_host(Cs) -> + case classify_host(Cs) of + regname -> Cs; + ipv4 -> Cs; + ipv6 -> bracket_ipv6(Cs); + other -> encode(Cs, fun is_reg_name/1) + end. + +-spec encode_path(list()|binary()) -> list() | binary(). +encode_path(Cs) -> + encode(Cs, fun is_path/1). + +-spec encode_query(list()|binary()) -> list() | binary(). +encode_query(Cs) -> + encode(Cs, fun is_query/1). + +-spec encode_fragment(list()|binary()) -> list() | binary(). +encode_fragment(Cs) -> + encode(Cs, fun is_fragment/1). + +%%------------------------------------------------------------------------- +%% Helper funtions for percent-decode +%%------------------------------------------------------------------------- +decode(<<$%,C0,C1,Cs/binary>>, Fun, Acc) -> + case is_hex_digit(C0) andalso is_hex_digit(C1) of + true -> + B = ?HEX2DEC(C0)*16+?HEX2DEC(C1), + decode(Cs, Fun, <<Acc/binary, B>>); + false -> throw({error,invalid_percent_encoding,<<$%,C0,C1>>}) + end; +decode(<<C,Cs/binary>>, Fun, Acc) -> + case Fun(C) of + true -> decode(Cs, Fun, <<Acc/binary, C>>); + false -> throw({error,invalid_percent_encoding,<<C,Cs/binary>>}) + end; +decode(<<>>, _Fun, Acc) -> + Acc. + +%% Check if char is allowed in host +-spec is_host(char()) -> boolean(). +is_host($:) -> true; +is_host(Char) -> is_unreserved(Char) orelse is_sub_delim(Char). + +%% Check if char is allowed in path +-spec is_path(char()) -> boolean(). +is_path($/) -> true; +is_path(Char) -> is_pchar(Char). + + +%%------------------------------------------------------------------------- +%% Helper functions for percent-encode +%%------------------------------------------------------------------------- +-spec encode(list()|binary(), fun()) -> list() | binary(). +encode(Component, Fun) when is_list(Component) -> + B = unicode:characters_to_binary(Component), + unicode:characters_to_list(encode(B, Fun, <<>>)); +encode(Component, Fun) when is_binary(Component) -> + encode(Component, Fun, <<>>). +%% +encode(<<Char/utf8, Rest/binary>>, Fun, Acc) -> + C = encode_codepoint_binary(Char, Fun), + encode(Rest, Fun, <<Acc/binary,C/binary>>); +encode(<<Char, Rest/binary>>, _Fun, _Acc) -> + throw({error,invalid_input,<<Char,Rest/binary>>}); +encode(<<>>, _Fun, Acc) -> + Acc. + + +-spec encode_codepoint_binary(integer(), fun()) -> binary(). +encode_codepoint_binary(C, Fun) -> + case Fun(C) of + false -> percent_encode_binary(C); + true -> <<C>> + end. + + +-spec percent_encode_binary(integer()) -> binary(). +percent_encode_binary(Code) -> + percent_encode_binary(<<Code/utf8>>, <<>>). + + +percent_encode_binary(<<A:4,B:4,Rest/binary>>, Acc) -> + percent_encode_binary(Rest, <<Acc/binary,$%,(?DEC2HEX(A)),(?DEC2HEX(B))>>); +percent_encode_binary(<<>>, Acc) -> + Acc. + + +%%------------------------------------------------------------------------- +%%------------------------------------------------------------------------- +validate_scheme([]) -> true; +validate_scheme([H|T]) -> + case is_scheme(H) of + true -> validate_scheme(T); + false -> false + end; +validate_scheme(<<>>) -> true; +validate_scheme(<<H, Rest/binary>>) -> + case is_scheme(H) of + true -> validate_scheme(Rest); + false -> false + end. + + +%%------------------------------------------------------------------------- +%% Classifies hostname into the following categories: +%% regname, ipv4 - address does not contain reserved characters to be +%% percent-encoded +%% ipv6 - address does not contain reserved characters but it shall be +%% encolsed in brackets +%% other - address shall be percent-encoded +%%------------------------------------------------------------------------- +classify_host([]) -> other; +classify_host(Addr) when is_binary(Addr) -> + A = unicode:characters_to_list(Addr), + classify_host_ipv6(A); +classify_host(Addr) -> + classify_host_ipv6(Addr). + +classify_host_ipv6(Addr) -> + case is_ipv6_address(Addr) of + true -> ipv6; + false -> classify_host_ipv4(Addr) + end. + +classify_host_ipv4(Addr) -> + case is_ipv4_address(Addr) of + true -> ipv4; + false -> classify_host_regname(Addr) + end. + +classify_host_regname([]) -> regname; +classify_host_regname([H|T]) -> + case is_reg_name(H) of + true -> classify_host_regname(T); + false -> other + end. + +is_ipv4_address(Addr) -> + case inet:parse_ipv4strict_address(Addr) of + {ok, _} -> true; + {error, _} -> false + end. + +is_ipv6_address(Addr) -> + case inet:parse_ipv6strict_address(Addr) of + {ok, _} -> true; + {error, _} -> false + end. + +bracket_ipv6(Addr) when is_binary(Addr) -> + concat(<<$[,Addr/binary>>,<<$]>>); +bracket_ipv6(Addr) when is_list(Addr) -> + [$[|Addr] ++ "]". + + +%%------------------------------------------------------------------------- +%% Helper funtions for recompose +%%------------------------------------------------------------------------- + +%%------------------------------------------------------------------------- +%% Checks if input Map has valid combination of fields that can be +%% recomposed into a URI. +%% +%% The implementation is based on a decision tree that fulfills the +%% following rules: +%% - 'path' shall always be present in the input map +%% URI = scheme ":" hier-part [ "?" query ] [ "#" fragment ] +%% hier-part = "//" authority path-abempty +%% / path-absolute +%% / path-rootless +%% / path-empty +%% - 'host' shall be present in the input map when 'path' starts with +%% two slashes ("//") +%% path = path-abempty ; begins with "/" or is empty +%% / path-absolute ; begins with "/" but not "//" +%% / path-noscheme ; begins with a non-colon segment +%% / path-rootless ; begins with a segment +%% / path-empty ; zero characters +%% path-abempty = *( "/" segment ) +%% segment = *pchar +%% - 'host' shall be present if userinfo or port is present in input map +%% authority = [ userinfo "@" ] host [ ":" port ] +%% - All fields shall be valid (scheme, userinfo, host, port, path, query +%% or fragment). +%%------------------------------------------------------------------------- +is_valid_map(#{path := Path} = Map) -> + ((starts_with_two_slash(Path) andalso is_valid_map_host(Map)) + orelse + (maps:is_key(userinfo, Map) andalso is_valid_map_host(Map)) + orelse + (maps:is_key(port, Map) andalso is_valid_map_host(Map)) + orelse + all_fields_valid(Map)); +is_valid_map(#{}) -> + false. + + +is_valid_map_host(Map) -> + maps:is_key(host, Map) andalso all_fields_valid(Map). + + +all_fields_valid(Map) -> + Fun = fun(scheme, _, Acc) -> Acc; + (userinfo, _, Acc) -> Acc; + (host, _, Acc) -> Acc; + (port, _, Acc) -> Acc; + (path, _, Acc) -> Acc; + (query, _, Acc) -> Acc; + (fragment, _, Acc) -> Acc; + (_, _, _) -> false + end, + maps:fold(Fun, true, Map). + + +starts_with_two_slash([$/,$/|_]) -> + true; +starts_with_two_slash(?STRING_REST("//", _)) -> + true; +starts_with_two_slash(_) -> false. + + +update_scheme(#{scheme := Scheme}, _) -> + add_colon_postfix(encode_scheme(Scheme)); +update_scheme(#{}, _) -> + empty. + + +update_userinfo(#{userinfo := Userinfo}, empty) -> + add_auth_prefix(encode_userinfo(Userinfo)); +update_userinfo(#{userinfo := Userinfo}, URI) -> + concat(URI,add_auth_prefix(encode_userinfo(Userinfo))); +update_userinfo(#{}, empty) -> + empty; +update_userinfo(#{}, URI) -> + URI. + + +update_host(#{host := Host}, empty) -> + add_auth_prefix(encode_host(Host)); +update_host(#{host := Host} = Map, URI) -> + concat(URI,add_host_prefix(Map, encode_host(Host))); +update_host(#{}, empty) -> + empty; +update_host(#{}, URI) -> + URI. + + +%% URI cannot be empty for ports. E.g. ":8080" is not a valid URI +update_port(#{port := undefined}, URI) -> + concat(URI, <<":">>); +update_port(#{port := Port}, URI) -> + concat(URI,add_colon(encode_port(Port))); +update_port(#{}, URI) -> + URI. + + +update_path(#{path := Path}, empty) -> + encode_path(Path); +update_path(#{path := Path}, URI) -> + concat(URI,encode_path(Path)); +update_path(#{}, empty) -> + empty; +update_path(#{}, URI) -> + URI. + + +update_query(#{query := Query}, empty) -> + encode_query(Query); +update_query(#{query := Query}, URI) -> + concat(URI,add_question_mark(encode_query(Query))); +update_query(#{}, empty) -> + empty; +update_query(#{}, URI) -> + URI. + + +update_fragment(#{fragment := Fragment}, empty) -> + add_hashmark(encode_fragment(Fragment)); +update_fragment(#{fragment := Fragment}, URI) -> + concat(URI,add_hashmark(encode_fragment(Fragment))); +update_fragment(#{}, empty) -> + ""; +update_fragment(#{}, URI) -> + URI. + +%%------------------------------------------------------------------------- +%% Concatenates its arguments that can be lists and binaries. +%% The result is a list if at least one of its argument is a list and +%% binary otherwise. +%%------------------------------------------------------------------------- +concat(A, B) when is_binary(A), is_binary(B) -> + <<A/binary, B/binary>>; +concat(A, B) when is_binary(A), is_list(B) -> + unicode:characters_to_list(A) ++ B; +concat(A, B) when is_list(A) -> + A ++ maybe_to_list(B). + +add_hashmark(Comp) when is_binary(Comp) -> + <<$#, Comp/binary>>; +add_hashmark(Comp) when is_list(Comp) -> + [$#|Comp]. + +add_question_mark(Comp) when is_binary(Comp) -> + <<$?, Comp/binary>>; +add_question_mark(Comp) when is_list(Comp) -> + [$?|Comp]. + +add_colon(Comp) when is_binary(Comp) -> + <<$:, Comp/binary>>. + +add_colon_postfix(Comp) when is_binary(Comp) -> + <<Comp/binary,$:>>; +add_colon_postfix(Comp) when is_list(Comp) -> + Comp ++ ":". + +add_auth_prefix(Comp) when is_binary(Comp) -> + <<"//", Comp/binary>>; +add_auth_prefix(Comp) when is_list(Comp) -> + [$/,$/|Comp]. + +add_host_prefix(#{userinfo := _}, Host) when is_binary(Host) -> + <<$@,Host/binary>>; +add_host_prefix(#{}, Host) when is_binary(Host) -> + <<"//",Host/binary>>; +add_host_prefix(#{userinfo := _}, Host) when is_list(Host) -> + [$@|Host]; +add_host_prefix(#{}, Host) when is_list(Host) -> + [$/,$/|Host]. + +maybe_to_list(Comp) when is_binary(Comp) -> unicode:characters_to_list(Comp); +maybe_to_list(Comp) -> Comp. + +encode_port(Port) -> + integer_to_binary(Port). + +%%------------------------------------------------------------------------- +%% Helper functions for transcode +%%------------------------------------------------------------------------- + +%%------------------------------------------------------------------------- +%% uri_string:transcode(<<"x%00%00%00%F6"/utf32>>). +%% 1. Convert (transcode/2) input to list form (list of unicode codepoints) +%% "x%00%00%00%F6" +%% 2. Accumulate characters until percent-encoded segment (transcode/4). +%% Acc = "x" +%% 3. Convert percent-encoded triplets to binary form (transcode_pct/4) +%% <<0,0,0,246>> +%% 4. Transcode in-encoded binary to out-encoding (utf32 -> utf8): +%% <<195,182>> +%% 5. Percent-encode out-encoded binary: +%% <<"%C3%B6"/utf8>> = <<37,67,51,37,66,54>> +%% 6. Convert binary to list form, reverse it and append the accumulator +%% "6B%3C%" + "x" +%% 7. Reverse Acc and return it +%%------------------------------------------------------------------------- +transcode([$%,_C0,_C1|_Rest] = L, Acc, InEnc, OutEnc) -> + transcode_pct(L, Acc, <<>>, InEnc, OutEnc); +transcode([_C|_Rest] = L, Acc, InEnc, OutEnc) -> + transcode(L, Acc, [], InEnc, OutEnc). +%% +transcode([$%,_C0,_C1|_Rest] = L, Acc, List, InEncoding, OutEncoding) -> + transcode_pct(L, List ++ Acc, <<>>, InEncoding, OutEncoding); +transcode([C|Rest], Acc, List, InEncoding, OutEncoding) -> + transcode(Rest, Acc, [C|List], InEncoding, OutEncoding); +transcode([], Acc, List, _InEncoding, _OutEncoding) -> + lists:reverse(List ++ Acc). + + +%% Transcode percent-encoded segment +transcode_pct([$%,C0,C1|Rest] = L, Acc, B, InEncoding, OutEncoding) -> + case is_hex_digit(C0) andalso is_hex_digit(C1) of + true -> + Int = ?HEX2DEC(C0)*16+?HEX2DEC(C1), + transcode_pct(Rest, Acc, <<B/binary, Int>>, InEncoding, OutEncoding); + false -> throw({error, invalid_percent_encoding,L}) + end; +transcode_pct([_C|_Rest] = L, Acc, B, InEncoding, OutEncoding) -> + OutBinary = convert_to_binary(B, InEncoding, OutEncoding), + PctEncUtf8 = percent_encode_segment(OutBinary), + Out = lists:reverse(convert_to_list(PctEncUtf8, utf8)), + transcode(L, Out ++ Acc, [], InEncoding, OutEncoding); +transcode_pct([], Acc, B, InEncoding, OutEncoding) -> + OutBinary = convert_to_binary(B, InEncoding, OutEncoding), + PctEncUtf8 = percent_encode_segment(OutBinary), + Out = convert_to_list(PctEncUtf8, utf8), + lists:reverse(Acc) ++ Out. + + +%% Convert to binary +convert_to_binary(Binary, InEncoding, OutEncoding) -> + case unicode:characters_to_binary(Binary, InEncoding, OutEncoding) of + {error, _List, RestData} -> + throw({error, invalid_input, RestData}); + {incomplete, _List, RestData} -> + throw({error, invalid_input, RestData}); + Result -> + Result + end. + + +%% Convert to list +convert_to_list(Binary, InEncoding) -> + case unicode:characters_to_list(Binary, InEncoding) of + {error, _List, RestData} -> + throw({error, invalid_input, RestData}); + {incomplete, _List, RestData} -> + throw({error, invalid_input, RestData}); + Result -> + Result + end. + + +%% Flatten input list +flatten_list([], _) -> + []; +flatten_list(L, InEnc) -> + flatten_list(L, InEnc, []). +%% +flatten_list([H|T], InEnc, Acc) when is_binary(H) -> + L = convert_to_list(H, InEnc), + flatten_list(T, InEnc, lists:reverse(L) ++ Acc); +flatten_list([H|T], InEnc, Acc) when is_list(H) -> + flatten_list(H ++ T, InEnc, Acc); +flatten_list([H|T], InEnc, Acc) -> + flatten_list(T, InEnc, [H|Acc]); +flatten_list([], _InEnc, Acc) -> + lists:reverse(Acc); +flatten_list(Arg, _, _) -> + throw({error, invalid_input, Arg}). + + +percent_encode_segment(Segment) -> + percent_encode_binary(Segment, <<>>). + + +%%------------------------------------------------------------------------- +%% Helper functions for normalize +%%------------------------------------------------------------------------- + +%% 6.2.2.1. Case Normalization +normalize_case(#{scheme := Scheme, host := Host} = Map) -> + Map#{scheme => to_lower(Scheme), + host => to_lower(Host)}; +normalize_case(#{host := Host} = Map) -> + Map#{host => to_lower(Host)}; +normalize_case(#{scheme := Scheme} = Map) -> + Map#{scheme => to_lower(Scheme)}; +normalize_case(#{} = Map) -> + Map. + + +to_lower(Cs) when is_list(Cs) -> + B = convert_to_binary(Cs, utf8, utf8), + convert_to_list(to_lower(B), utf8); +to_lower(Cs) when is_binary(Cs) -> + to_lower(Cs, <<>>). +%% +to_lower(<<C,Cs/binary>>, Acc) when $A =< C, C =< $Z -> + to_lower(Cs, <<Acc/binary,(C + 32)>>); +to_lower(<<C,Cs/binary>>, Acc) -> + to_lower(Cs, <<Acc/binary,C>>); +to_lower(<<>>, Acc) -> + Acc. + + +%% 6.2.2.3. Path Segment Normalization +%% 5.2.4. Remove Dot Segments +normalize_path_segment(Map) -> + Path = maps:get(path, Map, undefined), + Map#{path => remove_dot_segments(Path)}. + + +remove_dot_segments(Path) when is_binary(Path) -> + remove_dot_segments(Path, <<>>); +remove_dot_segments(Path) when is_list(Path) -> + B = convert_to_binary(Path, utf8, utf8), + B1 = remove_dot_segments(B, <<>>), + convert_to_list(B1, utf8). +%% +remove_dot_segments(<<>>, Output) -> + Output; +remove_dot_segments(<<"../",T/binary>>, Output) -> + remove_dot_segments(T, Output); +remove_dot_segments(<<"./",T/binary>>, Output) -> + remove_dot_segments(T, Output); +remove_dot_segments(<<"/./",T/binary>>, Output) -> + remove_dot_segments(<<$/,T/binary>>, Output); +remove_dot_segments(<<"/.">>, Output) -> + remove_dot_segments(<<$/>>, Output); +remove_dot_segments(<<"/../",T/binary>>, Output) -> + Out1 = remove_last_segment(Output), + remove_dot_segments(<<$/,T/binary>>, Out1); +remove_dot_segments(<<"/..">>, Output) -> + Out1 = remove_last_segment(Output), + remove_dot_segments(<<$/>>, Out1); +remove_dot_segments(<<$.>>, Output) -> + remove_dot_segments(<<>>, Output); +remove_dot_segments(<<"..">>, Output) -> + remove_dot_segments(<<>>, Output); +remove_dot_segments(Input, Output) -> + {First, Rest} = first_path_segment(Input), + remove_dot_segments(Rest, <<Output/binary,First/binary>>). + + +first_path_segment(Input) -> + F = first_path_segment(Input, <<>>), + split_binary(Input, byte_size(F)). +%% +first_path_segment(<<$/,T/binary>>, Acc) -> + first_path_segment_end(<<T/binary>>, <<Acc/binary,$/>>); +first_path_segment(<<C,T/binary>>, Acc) -> + first_path_segment_end(<<T/binary>>, <<Acc/binary,C>>). + + +first_path_segment_end(<<>>, Acc) -> + Acc; +first_path_segment_end(<<$/,_/binary>>, Acc) -> + Acc; +first_path_segment_end(<<C,T/binary>>, Acc) -> + first_path_segment_end(<<T/binary>>, <<Acc/binary,C>>). + + +remove_last_segment(<<>>) -> + <<>>; +remove_last_segment(B) -> + {Init, Last} = split_binary(B, byte_size(B) - 1), + case Last of + <<$/>> -> + Init; + _Char -> + remove_last_segment(Init) + end. + + +%% RFC 3986, 6.2.3. Scheme-Based Normalization +normalize_scheme_based(Map) -> + Scheme = maps:get(scheme, Map, undefined), + Port = maps:get(port, Map, undefined), + Path= maps:get(path, Map, undefined), + normalize_scheme_based(Map, Scheme, Port, Path). +%% +normalize_scheme_based(Map, Scheme, Port, Path) + when Scheme =:= "http"; Scheme =:= <<"http">> -> + normalize_http(Map, Port, Path); +normalize_scheme_based(Map, Scheme, Port, Path) + when Scheme =:= "https"; Scheme =:= <<"https">> -> + normalize_https(Map, Port, Path); +normalize_scheme_based(Map, Scheme, Port, _Path) + when Scheme =:= "ftp"; Scheme =:= <<"ftp">> -> + normalize_ftp(Map, Port); +normalize_scheme_based(Map, Scheme, Port, _Path) + when Scheme =:= "ssh"; Scheme =:= <<"ssh">> -> + normalize_ssh_sftp(Map, Port); +normalize_scheme_based(Map, Scheme, Port, _Path) + when Scheme =:= "sftp"; Scheme =:= <<"sftp">> -> + normalize_ssh_sftp(Map, Port); +normalize_scheme_based(Map, Scheme, Port, _Path) + when Scheme =:= "tftp"; Scheme =:= <<"tftp">> -> + normalize_tftp(Map, Port); +normalize_scheme_based(Map, _, _, _) -> + Map. + + +normalize_http(Map, Port, Path) -> + M1 = normalize_port(Map, Port, 80), + normalize_http_path(M1, Path). + + +normalize_https(Map, Port, Path) -> + M1 = normalize_port(Map, Port, 443), + normalize_http_path(M1, Path). + + +normalize_ftp(Map, Port) -> + normalize_port(Map, Port, 21). + + +normalize_ssh_sftp(Map, Port) -> + normalize_port(Map, Port, 22). + + +normalize_tftp(Map, Port) -> + normalize_port(Map, Port, 69). + + +normalize_port(Map, Port, Default) -> + case Port of + Default -> + maps:remove(port, Map); + _Else -> + Map + end. + + +normalize_http_path(Map, Path) -> + case Path of + "" -> + Map#{path => "/"}; + <<>> -> + Map#{path => <<"/">>}; + _Else -> + Map + end. diff --git a/lib/stdlib/test/Makefile b/lib/stdlib/test/Makefile index 523cb95065..8490770f3d 100644 --- a/lib/stdlib/test/Makefile +++ b/lib/stdlib/test/Makefile @@ -87,6 +87,7 @@ MODULES= \ timer_simple_SUITE \ unicode_SUITE \ unicode_util_SUITE \ + uri_string_SUITE \ win32reg_SUITE \ y2k_SUITE \ select_SUITE \ diff --git a/lib/stdlib/test/ets_SUITE.erl b/lib/stdlib/test/ets_SUITE.erl index 5a5e282998..f329a07b7a 100644 --- a/lib/stdlib/test/ets_SUITE.erl +++ b/lib/stdlib/test/ets_SUITE.erl @@ -1682,7 +1682,7 @@ do_random_test() -> ets:delete(Set), verify_etsmem(EtsMem). -%% Ttest various variants of update_element. +%% Test various variants of update_element. update_element(Config) when is_list(Config) -> EtsMem = etsmem(), repeat_for_opts(fun update_element_opts/1), diff --git a/lib/stdlib/test/property_test/README b/lib/stdlib/test/property_test/README new file mode 100644 index 0000000000..57602bf719 --- /dev/null +++ b/lib/stdlib/test/property_test/README @@ -0,0 +1,12 @@ + +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% +%%% %%% +%%% WARNING %%% +%%% %%% +%%% This is experimental code which may be changed or removed %%% +%%% anytime without any warning. %%% +%%% %%% +%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%% + +The test in this directory are written assuming that the user has a QuickCheck license. They are to be run manually. Some may be possible to be run with other tools, e.g. PropEr. + diff --git a/lib/stdlib/test/property_test/uri_string_recompose.erl b/lib/stdlib/test/property_test/uri_string_recompose.erl new file mode 100644 index 0000000000..e51a671172 --- /dev/null +++ b/lib/stdlib/test/property_test/uri_string_recompose.erl @@ -0,0 +1,361 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2008-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +-module(uri_string_recompose). + +-compile(export_all). + +-proptest(eqc). +-proptest([triq,proper]). + +-ifndef(EQC). +-ifndef(PROPER). +-ifndef(TRIQ). +-define(EQC,true). +-endif. +-endif. +-endif. + +-ifdef(EQC). +-include_lib("eqc/include/eqc.hrl"). +-define(MOD_eqc,eqc). + +-else. +-ifdef(PROPER). +-include_lib("proper/include/proper.hrl"). +-define(MOD_eqc,proper). + +-else. +-ifdef(TRIQ). +-define(MOD_eqc,triq). +-include_lib("triq/include/triq.hrl"). + +-endif. +-endif. +-endif. + + +-define(STRING_REST(MatchStr, Rest), <<MatchStr/utf8, Rest/binary>>). + +-define(SCHEME, {scheme, scheme()}). +-define(USER, {userinfo, unicode()}). +-define(HOST, {host, host_map()}). +-define(PORT, {port, port()}). +-define(PATH_ABE, {path, path_abempty_map()}). +-define(PATH_ABS, {path, path_absolute_map()}). +-define(PATH_NOS, {path, path_noscheme_map()}). +-define(PATH_ROO, {path, path_rootless_map()}). +-define(PATH_EMP, {path, path_empty_map()}). +-define(QUERY, {query, query_map()}). +-define(FRAGMENT, {fragment, fragment_map()}). + + +%%%======================================================================== +%%% Properties +%%%======================================================================== + +prop_recompose() -> + ?FORALL(Map, map(), + Map =:= uri_string:parse(uri_string:recompose(Map)) + ). + +%% Stats +prop_map_key_length_collect() -> + ?FORALL(List, map(), + collect(length(maps:keys(List)), true)). + +prop_map_collect() -> + ?FORALL(List, map(), + collect(lists:sort(maps:keys(List)), true)). + +prop_scheme_collect() -> + ?FORALL(List, scheme(), + collect(length(List), true)). + + +%%%======================================================================== +%%% Generators +%%%======================================================================== + +map() -> + ?LET(Gen, comp_proplist(), proplist_to_map(Gen)). + +comp_proplist() -> + frequency([ + {2, [?SCHEME,?PATH_ABS]}, + {2, [?SCHEME,?PATH_ROO]}, + {2, [?SCHEME,?PATH_EMP]}, + {2, [?SCHEME,?HOST,?PATH_ABE]}, + {2, [?SCHEME,?USER,?HOST,?PATH_ABE]}, + {2, [?SCHEME,?HOST,?PORT,?PATH_ABE]}, + {2, [?SCHEME,?USER,?HOST,?PORT,?PATH_ABE]}, + + {2, [?PATH_ABS]}, + {2, [?PATH_NOS]}, + {2, [?PATH_EMP]}, + {2, [?HOST,?PATH_ABE]}, + {2, [?USER,?HOST,?PATH_ABE]}, + {2, [?HOST,?PORT,?PATH_ABE]}, + {2, [?USER,?HOST,?PORT,?PATH_ABE]}, + + + {2, [?SCHEME,?PATH_ABS,?QUERY]}, + {2, [?SCHEME,?PATH_ROO,?QUERY]}, + {2, [?SCHEME,?PATH_EMP,?QUERY]}, + {2, [?SCHEME,?HOST,?PATH_ABE,?QUERY]}, + {2, [?SCHEME,?USER,?HOST,?PATH_ABE,?QUERY]}, + {2, [?SCHEME,?HOST,?PORT,?PATH_ABE,?QUERY]}, + {2, [?SCHEME,?USER,?HOST,?PORT,?PATH_ABE,?QUERY]}, + + {2, [?PATH_ABS,?QUERY]}, + {2, [?PATH_NOS,?QUERY]}, + {2, [?PATH_EMP,?QUERY]}, + {2, [?HOST,?PATH_ABE,?QUERY]}, + {2, [?USER,?HOST,?PATH_ABE,?QUERY]}, + {2, [?HOST,?PORT,?PATH_ABE,?QUERY]}, + {2, [?USER,?HOST,?PORT,?PATH_ABE,?QUERY]}, + + + {2, [?SCHEME,?PATH_ABS,?FRAGMENT]}, + {2, [?SCHEME,?PATH_ROO,?FRAGMENT]}, + {2, [?SCHEME,?PATH_EMP,?FRAGMENT]}, + {2, [?SCHEME,?HOST,?PATH_ABE,?FRAGMENT]}, + {2, [?SCHEME,?USER,?HOST,?PATH_ABE,?FRAGMENT]}, + {2, [?SCHEME,?HOST,?PORT,?PATH_ABE,?FRAGMENT]}, + {2, [?SCHEME,?USER,?HOST,?PORT,?PATH_ABE,?FRAGMENT]}, + + {2, [?PATH_ABS,?FRAGMENT]}, + {2, [?PATH_NOS,?FRAGMENT]}, + {2, [?PATH_EMP,?FRAGMENT]}, + {2, [?HOST,?PATH_ABE,?FRAGMENT]}, + {2, [?USER,?HOST,?PATH_ABE,?FRAGMENT]}, + {2, [?HOST,?PORT,?PATH_ABE,?FRAGMENT]}, + {2, [?USER,?HOST,?PORT,?PATH_ABE,?FRAGMENT]}, + + + {2, [?SCHEME,?PATH_ABS,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?PATH_ROO,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?PATH_EMP,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?HOST,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?USER,?HOST,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?HOST,?PORT,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?SCHEME,?USER,?HOST,?PORT,?PATH_ABE,?QUERY,?FRAGMENT]}, + + {2, [?PATH_ABS,?QUERY,?FRAGMENT]}, + {2, [?PATH_NOS,?QUERY,?FRAGMENT]}, + {2, [?PATH_EMP,?QUERY,?FRAGMENT]}, + {2, [?HOST,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?USER,?HOST,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?HOST,?PORT,?PATH_ABE,?QUERY,?FRAGMENT]}, + {2, [?USER,?HOST,?PORT,?PATH_ABE,?QUERY,?FRAGMENT]} + ]). + + +%%------------------------------------------------------------------------- +%% Path +%%------------------------------------------------------------------------- +path_abempty_map() -> + frequency([{90, path_abe_map()}, + {10, path_empty_map()}]). + +path_abe_map() -> + ?SIZED(Length, path_abe_map(Length, [])). +%% +path_abe_map(0, Segments) -> + ?LET(Gen, Segments, lists:append(Gen)); +path_abe_map(N, Segments) -> + path_abe_map(N-1, [slash(),segment()|Segments]). + + +path_absolute_map() -> + ?SIZED(Length, path_absolute_map(Length, [])). +%% +path_absolute_map(0, Segments) -> + ?LET(Gen, [slash(),segment_nz()|Segments], lists:append(Gen)); +path_absolute_map(N, Segments) -> + path_absolute_map(N-1, [slash(),segment()|Segments]). + + +path_noscheme_map() -> + ?SIZED(Length, path_noscheme_map(Length, [])). +%% +path_noscheme_map(0, Segments) -> + ?LET(Gen, [segment_nz_nc()|Segments], lists:append(Gen)); +path_noscheme_map(N, Segments) -> + path_noscheme_map(N-1, [slash(),segment()|Segments]). + +path_rootless_map() -> + ?SIZED(Length, path_rootless_map(Length, [])). +%% +path_rootless_map(0, Segments) -> + ?LET(Gen, [segment_nz()|Segments], lists:append(Gen)); +path_rootless_map(N, Segments) -> + path_rootless_map(N-1, [slash(),segment()|Segments]). + + +segment_nz() -> + non_empty(segment()). + +segment_nz_nc() -> + non_empty(list(frequency([{30, unreserved()}, + {10, sub_delims()}, + {10, unicode_char()}, + {5, oneof([$@])} + ]))). + + +segment() -> + list(frequency([{30, unreserved()}, + {10, sub_delims()}, + {10, unicode_char()}, + {5, oneof([$:, $@])} + ])). + +slash() -> + "/". + +path_empty_map() -> + "". + + +%%------------------------------------------------------------------------- +%% Path +%%------------------------------------------------------------------------- +host_map() -> + frequency([{30, reg_name()}, + {30, ip_address()} + ]). + + +reg_name() -> + list(frequency([{30, alpha()}, + {10, sub_delims()}, + {10, unicode_char()} + ])). + +ip_address() -> + oneof(["127.0.0.1", "::127.0.0.1", + "2001:0db8:0000:0000:0000:0000:1428:07ab", + "2001:0db8:0000:0000:0000::1428:07ab", + "2001:0db8:0:0:0:0:1428:07ab", + "2001:0db8:0::0:1428:07ab"]). + +%% Generating only reg-names +host_uri() -> + non_empty(list(frequency([{30, unreserved()}, + {10, sub_delims()}, + {10, pct_encoded()} + ]))). + +%%------------------------------------------------------------------------- +%% Port, Query, Fragment +%%------------------------------------------------------------------------- +port() -> + frequency([{10, undefined}, + {10, range(1,65535)} + ]). + +query_map() -> + unicode(). + + +query_uri() -> + [$?| non_empty(list(frequency([{20, pchar()}, + {5, oneof([$/, $?])} % punctuation + ])))]. + +fragment_map() -> + unicode(). + +fragment_uri() -> + [$?| non_empty(list(frequency([{20, pchar()}, + {5, oneof([$/, $?])} % punctuation + ])))]. + + +%%------------------------------------------------------------------------- +%% Scheme +%%------------------------------------------------------------------------- +scheme() -> + ?SIZED(Length, scheme_start(Length, [])). +%% +scheme_start(0, L) -> + ?LET(Gen, L, lists:reverse(Gen)); +scheme_start(N, L) -> + scheme(N-1,[alpha()|L]). + +scheme(0, L) -> + ?LET(Gen, L, lists:reverse(Gen)); +scheme(N, L) -> + scheme(N-1, [scheme_char()|L]). + + +%%------------------------------------------------------------------------- +%% Misc +%%------------------------------------------------------------------------- +unicode() -> + list(frequency([{20, alpha()}, % alpha + {10, digit()}, % digit + {10, unicode_char()} % unicode + ])). + +scheme_char() -> + frequency([{20, alpha()}, % alpha + {20, digit()}, % digit + {5, oneof([$+, $-, $.])} % punctuation + ]). + +sub_delims() -> + oneof([$!, $$, $&, $', $(, $), + $*, $+, $,,$;, $=]). + +pchar() -> + frequency([{20, unreserved()}, + {5, pct_encoded()}, + {5, sub_delims()}, + {1, oneof([$:, $@])} % punctuation + ]). + +unreserved() -> + frequency([{20, alpha()}, + {5, digit()}, + {1, oneof([$-, $., $_, $~])} % punctuation + ]). + +unicode_char() -> + range(913, 1023). + +alpha() -> + frequency([{20, range($a, $z)}, % letters + {20, range($A, $Z)}]). % letters + +digit() -> + range($0, $9). % numbers + +pct_encoded() -> + oneof(["%C3%A4", "%C3%A5", "%C3%B6"]). + + +%%%======================================================================== +%%% Helpers +%%%======================================================================== +proplist_to_map(L) -> + lists:foldl(fun({K,V},M) -> M#{K => V}; + (_,M) -> M + end, #{}, L). diff --git a/lib/stdlib/test/uri_string_SUITE.erl b/lib/stdlib/test/uri_string_SUITE.erl new file mode 100644 index 0000000000..c625da56c6 --- /dev/null +++ b/lib/stdlib/test/uri_string_SUITE.erl @@ -0,0 +1,844 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2008-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +-module(uri_string_SUITE). + +-include_lib("common_test/include/ct.hrl"). + +-export([all/0, suite/0,groups/0, + normalize/1, + parse_binary_fragment/1, parse_binary_host/1, parse_binary_host_ipv4/1, + parse_binary_host_ipv6/1, + parse_binary_path/1, parse_binary_pct_encoded_fragment/1, parse_binary_pct_encoded_query/1, + parse_binary_pct_encoded_userinfo/1, parse_binary_port/1, + parse_binary_query/1, parse_binary_scheme/1, parse_binary_userinfo/1, + parse_fragment/1, parse_host/1, parse_host_ipv4/1, parse_host_ipv6/1, + parse_path/1, parse_pct_encoded_fragment/1, parse_pct_encoded_query/1, + parse_pct_encoded_userinfo/1, parse_port/1, + parse_query/1, parse_scheme/1, parse_userinfo/1, + parse_list/1, parse_binary/1, parse_mixed/1, parse_relative/1, + parse_special/1, parse_special2/1, parse_negative/1, + recompose_fragment/1, recompose_parse_fragment/1, + recompose_query/1, recompose_parse_query/1, + recompose_path/1, recompose_parse_path/1, + recompose_autogen/1, parse_recompose_autogen/1, + transcode_basic/1, transcode_options/1, transcode_mixed/1, transcode_negative/1 + ]). + + +-define(SCHEME, "foo"). +-define(USERINFO, "åsa"). +-define(USERINFO_ENC, "%C3%A5sa"). +-define(HOST, "älvsjö"). +-define(HOST_ENC, "%C3%A4lvsj%C3%B6"). +-define(IPV6, "::127.0.0.1"). +-define(IPV6_ENC, "[::127.0.0.1]"). +-define(PORT, 8042). +-define(PORT_ENC, ":8042"). +-define(PATH, "/där"). +-define(PATH_ENC, "/d%C3%A4r"). +-define(QUERY, "name=örn"). +-define(QUERY_ENC, "?name=%C3%B6rn"). +-define(FRAGMENT, "näsa"). +-define(FRAGMENT_ENC, "#n%C3%A4sa"). + + +suite() -> + [{timetrap,{minutes,1}}]. + +all() -> + [ + normalize, + parse_binary_scheme, + parse_binary_userinfo, + parse_binary_pct_encoded_userinfo, + parse_binary_host, + parse_binary_host_ipv4, + parse_binary_host_ipv6, + parse_binary_port, + parse_binary_path, + parse_binary_query, + parse_binary_pct_encoded_query, + parse_binary_fragment, + parse_binary_pct_encoded_fragment, + parse_scheme, + parse_userinfo, + parse_pct_encoded_userinfo, + parse_host, + parse_host_ipv4, + parse_host_ipv6, + parse_port, + parse_path, + parse_query, + parse_pct_encoded_query, + parse_fragment, + parse_pct_encoded_fragment, + parse_list, + parse_binary, + parse_mixed, + parse_relative, + parse_special, + parse_special2, + parse_negative, + recompose_fragment, + recompose_parse_fragment, + recompose_query, + recompose_parse_query, + recompose_path, + recompose_parse_path, + recompose_autogen, + parse_recompose_autogen, + transcode_basic, + transcode_options, + transcode_mixed, + transcode_negative + ]. + +groups() -> + []. + + +%%------------------------------------------------------------------------- +%% Helper functions +%%------------------------------------------------------------------------- +uri_combinations() -> + [[Sch,Usr,Hst,Prt,Pat,Qry,Frg] || + Sch <- [fun update_scheme/1, fun update_scheme_binary/1, none], + Usr <- [fun update_userinfo/1, fun update_userinfo_binary/1, none], + Hst <- [fun update_host/1, fun update_host_binary/1, + fun update_ipv6/1, fun update_ipv6_binary/1, none], + Prt <- [fun update_port/1, none], + Pat <- [fun update_path/1, fun update_path_binary/1], + Qry <- [fun update_query/1,fun update_query_binary/1, none], + Frg <- [fun update_fragment/1, fun update_fragment_binary/1, none], + not (Usr =:= none andalso Hst =:= none andalso Prt =/= none), + not (Usr =/= none andalso Hst =:= none andalso Prt =:= none), + not (Usr =/= none andalso Hst =:= none andalso Prt =/= none)]. + + +generate_test_vector(Comb) -> + Fun = fun (F, {Map, URI}) when is_function(F) -> F({Map, URI}); + (_, Map) -> Map + end, + lists:foldl(Fun, {#{}, empty}, Comb). + +generate_test_vectors(L) -> + lists:map(fun generate_test_vector/1, L). + +update_fragment({In, empty}) -> + {In#{fragment => ?FRAGMENT}, ?FRAGMENT_ENC}; +update_fragment({In, Out}) when is_list(Out) -> + {In#{fragment => ?FRAGMENT}, Out ++ ?FRAGMENT_ENC}; +update_fragment({In, Out}) when is_binary(Out) -> + {In#{fragment => ?FRAGMENT}, binary_to_list(Out) ++ ?FRAGMENT_ENC}. + +update_fragment_binary({In, empty}) -> + {In#{fragment => <<?FRAGMENT/utf8>>}, <<?FRAGMENT_ENC>>}; +update_fragment_binary({In, Out}) when is_list(Out) -> + {In#{fragment => <<?FRAGMENT/utf8>>}, Out ++ ?FRAGMENT_ENC}; +update_fragment_binary({In, Out}) when is_binary(Out) -> + {In#{fragment => <<?FRAGMENT/utf8>>}, <<Out/binary,?FRAGMENT_ENC>>}. + + +update_query({In, empty}) -> + {In#{query => ?QUERY}, ?QUERY_ENC}; +update_query({In, Out}) when is_list(Out) -> + {In#{query => ?QUERY}, Out ++ ?QUERY_ENC}; +update_query({In, Out}) when is_binary(Out) -> + {In#{query => ?QUERY}, binary_to_list(Out) ++ ?QUERY_ENC}. + +update_query_binary({In, empty}) -> + {In#{query => <<?QUERY/utf8>>}, <<?QUERY_ENC>>}; +update_query_binary({In, Out}) when is_list(Out) -> + {In#{query => <<?QUERY/utf8>>}, Out ++ ?QUERY_ENC}; +update_query_binary({In, Out}) when is_binary(Out) -> + {In#{query => <<?QUERY/utf8>>}, <<Out/binary,?QUERY_ENC>>}. + +update_path({In, empty}) -> + {In#{path => ?PATH}, ?PATH_ENC}; +update_path({In, Out}) when is_list(Out) -> + {In#{path => ?PATH}, Out ++ ?PATH_ENC}; +update_path({In, Out}) when is_binary(Out) -> + {In#{path => ?PATH}, binary_to_list(Out) ++ ?PATH_ENC}. + +update_path_binary({In, empty}) -> + {In#{path => <<?PATH/utf8>>}, <<?PATH_ENC>>}; +update_path_binary({In, Out}) when is_list(Out) -> + {In#{path => <<?PATH/utf8>>}, Out ++ ?PATH_ENC}; +update_path_binary({In, Out}) when is_binary(Out) -> + {In#{path => <<?PATH/utf8>>}, <<Out/binary,?PATH_ENC>>}. + +update_port({In, Out}) when is_list(Out) -> + {In#{port => ?PORT}, Out ++ ?PORT_ENC}; +update_port({In, Out}) when is_binary(Out) -> + {In#{port => ?PORT}, <<Out/binary,?PORT_ENC>>}. + +update_host({In, empty}) -> + {In#{host => ?HOST}, "//" ++ ?HOST_ENC}; +update_host({In, Out}) when is_list(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => ?HOST}, Out ++ [$@|?HOST_ENC]}; + false -> {In#{host => ?HOST}, Out ++ [$/,$/|?HOST_ENC]} + end; +update_host({In, Out}) when is_binary(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => ?HOST}, binary_to_list(Out) ++ [$@|?HOST_ENC]}; + false -> {In#{host => ?HOST}, binary_to_list(Out) ++ [$/,$/|?HOST_ENC]} + end. + +update_host_binary({In, empty}) -> + {In#{host => <<?HOST/utf8>>}, <<"//",?HOST_ENC>>}; +update_host_binary({In, Out}) when is_list(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => <<?HOST/utf8>>}, Out ++ [$@|?HOST_ENC]}; + false -> {In#{host => <<?HOST/utf8>>}, Out ++ [$/,$/|?HOST_ENC]} + end; +update_host_binary({In, Out}) when is_binary(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => <<?HOST/utf8>>}, <<Out/binary,$@,?HOST_ENC>>}; + false-> {In#{host => <<?HOST/utf8>>}, <<Out/binary,"//",?HOST_ENC>>} + end. + +update_ipv6({In, empty}) -> + {In#{host => ?IPV6}, "//" ++ ?IPV6_ENC}; +update_ipv6({In, Out}) when is_list(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => ?IPV6}, Out ++ [$@|?IPV6_ENC]}; + false -> {In#{host => ?IPV6}, Out ++ [$/,$/|?IPV6_ENC]} + end; +update_ipv6({In, Out}) when is_binary(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => ?IPV6}, binary_to_list(Out) ++ [$@|?IPV6_ENC]}; + false -> {In#{host => ?IPV6}, binary_to_list(Out) ++ [$/,$/|?IPV6_ENC]} + end. + +update_ipv6_binary({In, empty}) -> + {In#{host => <<?IPV6/utf8>>}, <<"//",?IPV6_ENC>>}; +update_ipv6_binary({In, Out}) when is_list(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => <<?IPV6/utf8>>}, Out ++ [$@|?IPV6_ENC]}; + false -> {In#{host => <<?IPV6/utf8>>}, Out ++ [$/,$/|?IPV6_ENC]} + end; +update_ipv6_binary({In, Out}) when is_binary(Out) -> + case maps:is_key(userinfo, In) of + true -> {In#{host => <<?IPV6/utf8>>}, <<Out/binary,$@,?IPV6_ENC>>}; + false-> {In#{host => <<?IPV6/utf8>>}, <<Out/binary,"//",?IPV6_ENC>>} + end. + +update_userinfo({In, empty}) -> + {In#{userinfo => ?USERINFO}, "//" ++ ?USERINFO_ENC}; +update_userinfo({In, Out}) when is_list(Out) -> + {In#{userinfo => ?USERINFO}, Out ++ "//" ++ ?USERINFO_ENC}; +update_userinfo({In, Out}) when is_binary(Out) -> + {In#{userinfo => ?USERINFO}, binary_to_list(Out) ++ "//" ++ ?USERINFO_ENC}. + +update_userinfo_binary({In, empty}) -> + {In#{userinfo => <<?USERINFO/utf8>>}, <<"//",?USERINFO_ENC>>}; +update_userinfo_binary({In, Out}) when is_list(Out) -> + {In#{userinfo => <<?USERINFO/utf8>>}, Out ++ "//" ++ ?USERINFO_ENC}; +update_userinfo_binary({In, Out}) when is_binary(Out) -> + {In#{userinfo => <<?USERINFO/utf8>>}, <<Out/binary,"//",?USERINFO_ENC>>}. + +update_scheme({In, empty}) -> + {In#{scheme => ?SCHEME}, ?SCHEME ++ ":"}. + +update_scheme_binary({In, empty}) -> + {In#{scheme => <<?SCHEME/utf8>>}, <<?SCHEME,$:>>}. + + +%% Test recompose on a generated test vector +run_test_recompose({#{}, empty}) -> + try "" = uri_string:recompose(#{}) of + _ -> ok + catch + _:_ -> error({test_failed, #{}, ""}) + end; +run_test_recompose({Map, URI}) -> + try URI = uri_string:recompose(Map) of + URI -> ok + catch + _:_ -> error({test_failed, Map, URI}) + end. + +%% Test parse - recompose on a generated test vector +run_test_parse_recompose({#{}, empty}) -> + try "" = uri_string:recompose(uri_string:parse("")) of + _ -> ok + catch + _:_ -> error({test_failed, #{}, ""}) + end; +run_test_parse_recompose({Map, URI}) -> + try URI = uri_string:recompose(uri_string:parse(URI)) of + URI -> ok + catch + _:_ -> error({test_failed, Map, URI}) + end. + + +%%------------------------------------------------------------------------- +%% Parse tests +%%------------------------------------------------------------------------- + +parse_binary_scheme(_Config) -> + #{} = uri_string:parse(<<>>), + #{path := <<"foo">>} = uri_string:parse(<<"foo">>), + #{scheme := <<"foo">>} = uri_string:parse(<<"foo:">>), + #{scheme := <<"foo">>, path := <<"bar:nisse">>} = uri_string:parse(<<"foo:bar:nisse">>), + #{scheme := <<"foo">>, host := <<"">>} = uri_string:parse(<<"foo://">>), + #{scheme := <<"foo">>, host := <<"">>, path := <<"/">>} = uri_string:parse(<<"foo:///">>), + #{scheme := <<"foo">>, host := <<"">>, path := <<"//">>} = uri_string:parse(<<"foo:////">>), + + #{path := <<"/">>} = uri_string:parse(<<"/">>), + #{host := <<>>} = uri_string:parse(<<"//">>), + #{host := <<>>, path := <<"/">>} = uri_string:parse(<<"///">>). + +parse_binary_userinfo(_Config) -> + #{scheme := <<"user">>, path := <<"password@localhost">>} = + uri_string:parse(<<"user:password@localhost">>), + #{path := <<"user@">>} = uri_string:parse(<<"user@">>), + #{path := <<"/user@">>} = uri_string:parse(<<"/user@">>), + #{path := <<"user@localhost">>} = uri_string:parse(<<"user@localhost">>), + #{userinfo := <<"user">>, host := <<"localhost">>} = uri_string:parse(<<"//user@localhost">>), + #{userinfo := <<"user:password">>, host := <<"localhost">>} = + uri_string:parse(<<"//user:password@localhost">>), + #{scheme := <<"foo">>, path := <<"/user@">>} = + uri_string:parse(<<"foo:/user@">>), + #{scheme := <<"foo">>, userinfo := <<"user">>, host := <<"localhost">>} = + uri_string:parse(<<"foo://user@localhost">>), + #{scheme := <<"foo">>, userinfo := <<"user:password">>, host := <<"localhost">>} = + uri_string:parse(<<"foo://user:password@localhost">>). + +parse_binary_pct_encoded_userinfo(_Config) -> + #{scheme := <<"user">>, path := <<"合@気道"/utf8>>} = + uri_string:parse(<<"user:%E5%90%88@%E6%B0%97%E9%81%93">>), + #{path := <<"合気道@"/utf8>>} = uri_string:parse(<<"%E5%90%88%E6%B0%97%E9%81%93@">>), + #{path := <<"/合気道@"/utf8>>} = uri_string:parse(<<"/%E5%90%88%E6%B0%97%E9%81%93@">>), + #{path := <<"合@気道"/utf8>>} = uri_string:parse(<<"%E5%90%88@%E6%B0%97%E9%81%93">>), + #{userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + uri_string:parse(<<"//%E5%90%88@%E6%B0%97%E9%81%93">>), + #{userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + uri_string:parse(<<"//%E5%90%88:%E6%B0%97@%E9%81%93">>), + #{scheme := <<"foo">>, path := <<"/合気道@"/utf8>>} = + uri_string:parse(<<"foo:/%E5%90%88%E6%B0%97%E9%81%93@">>), + #{scheme := <<"foo">>, userinfo := <<"合"/utf8>>, host := <<"気道"/utf8>>} = + uri_string:parse(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93">>), + #{scheme := <<"foo">>, userinfo := <<"合:気"/utf8>>, host := <<"道"/utf8>>} = + uri_string:parse(<<"foo://%E5%90%88:%E6%B0%97@%E9%81%93">>), + {error,invalid_uri,"@"} = uri_string:parse(<<"//%E5%90%88@%E6%B0%97%E9%81%93@">>), + {error,invalid_uri,":"} = uri_string:parse(<<"foo://%E5%90%88@%E6%B0%97%E9%81%93@">>). + +parse_binary_host(_Config) -> + #{host := <<"hostname">>} = uri_string:parse(<<"//hostname">>), + #{host := <<"hostname">>,scheme := <<"foo">>} = uri_string:parse(<<"foo://hostname">>), + #{host := <<"hostname">>,scheme := <<"foo">>, userinfo := <<"user">>} = + uri_string:parse(<<"foo://user@hostname">>). + +parse_binary_host_ipv4(_Config) -> + #{host := <<"127.0.0.1">>} = uri_string:parse(<<"//127.0.0.1">>), + #{host := <<"127.0.0.1">>, path := <<"/over/there">>} = + uri_string:parse(<<"//127.0.0.1/over/there">>), + #{host := <<"127.0.0.1">>, query := <<"name=ferret">>} = + uri_string:parse(<<"//127.0.0.1?name=ferret">>), + #{host := <<"127.0.0.1">>, fragment := <<"nose">>} = uri_string:parse(<<"//127.0.0.1#nose">>), + {error,invalid_uri,"x"} = uri_string:parse(<<"//127.0.0.x">>), + {error,invalid_uri,"1227.0.0.1"} = uri_string:parse(<<"//1227.0.0.1">>). + +parse_binary_host_ipv6(_Config) -> + #{host := <<"::127.0.0.1">>} = uri_string:parse(<<"//[::127.0.0.1]">>), + #{host := <<"2001:0db8:0000:0000:0000:0000:1428:07ab">>} = + uri_string:parse(<<"//[2001:0db8:0000:0000:0000:0000:1428:07ab]">>), + #{host := <<"::127.0.0.1">>, path := <<"/over/there">>} = + uri_string:parse(<<"//[::127.0.0.1]/over/there">>), + #{host := <<"::127.0.0.1">>, query := <<"name=ferret">>} = + uri_string:parse(<<"//[::127.0.0.1]?name=ferret">>), + #{host := <<"::127.0.0.1">>, fragment := <<"nose">>} = + uri_string:parse(<<"//[::127.0.0.1]#nose">>), + {error,invalid_uri,"x"} = uri_string:parse(<<"//[::127.0.0.x]">>), + {error,invalid_uri,"::1227.0.0.1"} = uri_string:parse(<<"//[::1227.0.0.1]">>), + {error,invalid_uri,"G"} = uri_string:parse(<<"//[2001:0db8:0000:0000:0000:0000:1428:G7ab]">>). + +parse_binary_port(_Config) -> + #{path:= <<"/:8042">>} = + uri_string:parse(<<"/:8042">>), + #{host:= <<>>, port := 8042} = + uri_string:parse(<<"//:8042">>), + #{host := <<"example.com">>, port:= 8042} = + uri_string:parse(<<"//example.com:8042">>), + #{scheme := <<"foo">>, path := <<"/:8042">>} = + uri_string:parse(<<"foo:/:8042">>), + #{scheme := <<"foo">>, host := <<>>, port := 8042} = + uri_string:parse(<<"foo://:8042">>), + #{scheme := <<"foo">>, host := <<"example.com">>, port := 8042} = + uri_string:parse(<<"foo://example.com:8042">>), + {error,invalid_uri,":"} = uri_string:parse(":600"), + {error,invalid_uri,"x"} = uri_string:parse("//:8042x"). + +parse_binary_path(_Config) -> + #{path := <<"over/there">>} = uri_string:parse(<<"over/there">>), + #{path := <<"/over/there">>} = uri_string:parse(<<"/over/there">>), + #{scheme := <<"foo">>, path := <<"/over/there">>} = + uri_string:parse(<<"foo:/over/there">>), + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/over/there">>} = + uri_string:parse(<<"foo://example.com/over/there">>), + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/over/there">>, port := 8042} = + uri_string:parse(<<"foo://example.com:8042/over/there">>). + +parse_binary_query(_Config) -> + #{scheme := <<"foo">>, query := <<"name=ferret">>} = + uri_string:parse(<<"foo:?name=ferret">>), + #{scheme := <<"foo">>, path:= <<"over/there">>, query := <<"name=ferret">>} = + uri_string:parse(<<"foo:over/there?name=ferret">>), + #{scheme := <<"foo">>, path:= <<"/over/there">>, query := <<"name=ferret">>} = + uri_string:parse(<<"foo:/over/there?name=ferret">>), + #{scheme := <<"foo">>, host := <<"example.com">>, query := <<"name=ferret">>} = + uri_string:parse(<<"foo://example.com?name=ferret">>), + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/">>, query := <<"name=ferret">>} = + uri_string:parse(<<"foo://example.com/?name=ferret">>), + + #{path := <<>>, query := <<"name=ferret">>} = + uri_string:parse(<<"?name=ferret">>), + #{path := <<"over/there">>, query := <<"name=ferret">>} = + uri_string:parse(<<"over/there?name=ferret">>), + #{path := <<"/">>, query := <<"name=ferret">>} = + uri_string:parse(<<"/?name=ferret">>), + #{path := <<"/over/there">>, query := <<"name=ferret">>} = + uri_string:parse(<<"/over/there?name=ferret">>), + #{host := <<"example.com">>, query := <<"name=ferret">>} = + uri_string:parse(<<"//example.com?name=ferret">>), + #{host := <<"example.com">>, path := <<"/">>, query := <<"name=ferret">>} = + uri_string:parse(<<"//example.com/?name=ferret">>). + +parse_binary_pct_encoded_query(_Config) -> + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/">>, + query := <<"name=合気道"/utf8>>} = + uri_string:parse(<<"foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>), + #{host := <<"example.com">>, path := <<"/">>, query := <<"name=合気道"/utf8>>} = + uri_string:parse(<<"//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93">>). + +parse_binary_fragment(_Config) -> + #{scheme := <<"foo">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo:#nose">>), + #{scheme := <<"foo">>, path:= <<"over/there">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo:over/there#nose">>), + #{scheme := <<"foo">>, path:= <<"/over/there">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo:/over/there#nose">>), + #{scheme := <<"foo">>, host := <<"example.com">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo://example.com#nose">>), + #{scheme := <<"foo">>, host := <<"example.com">>, path := <<"/">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo://example.com/#nose">>), + #{scheme := <<"foo">>, host := <<"example.com">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo://example.com#nose">>), + + #{fragment := <<"nose">>} = + uri_string:parse(<<"#nose">>), + #{path := <<"over/there">>, fragment := <<"nose">>} = + uri_string:parse(<<"over/there#nose">>), + #{path := <<"/">>, fragment := <<"nose">>} = + uri_string:parse(<<"/#nose">>), + #{path := <<"/over/there">>, fragment := <<"nose">>} = + uri_string:parse(<<"/over/there#nose">>), + #{host := <<"example.com">>, fragment := <<"nose">>} = + uri_string:parse(<<"//example.com#nose">>), + #{host := <<"example.com">>, path := <<"/">>, fragment := <<"nose">>} = + uri_string:parse(<<"//example.com/#nose">>). + +parse_binary_pct_encoded_fragment(_Config) -> + #{scheme := <<"foo">>, host := <<"example.com">>, fragment := <<"合気道"/utf8>>} = + uri_string:parse(<<"foo://example.com#%E5%90%88%E6%B0%97%E9%81%93">>), + #{host := <<"example.com">>, path := <<"/">>, fragment := <<"合気道"/utf8>>} = + uri_string:parse(<<"//example.com/#%E5%90%88%E6%B0%97%E9%81%93">>). + +parse_scheme(_Config) -> + #{} = uri_string:parse(""), + #{path := "foo"} = uri_string:parse("foo"), + #{scheme := "foo"} = uri_string:parse("foo:"), + #{scheme := "foo", path := "bar:nisse"} = uri_string:parse("foo:bar:nisse"), + #{scheme := "foo", host := ""} = uri_string:parse("foo://"), + #{scheme := "foo", host := "", path := "/"} = uri_string:parse("foo:///"), + #{scheme := "foo", host := "", path := "//"} = uri_string:parse("foo:////"), + + #{path := "/"} = uri_string:parse("/"), + #{host := ""} = uri_string:parse("//"), + #{host := "", path := "/"} = uri_string:parse("///"). + +parse_userinfo(_Config) -> + #{scheme := "user", path := "password@localhost"} = uri_string:parse("user:password@localhost"), + #{path := "user@"} = uri_string:parse("user@"), + #{path := "/user@"} = uri_string:parse("/user@"), + #{path := "user@localhost"} = uri_string:parse("user@localhost"), + #{userinfo := "user", host := "localhost"} = uri_string:parse("//user@localhost"), + #{userinfo := "user:password", host := "localhost"} = + uri_string:parse("//user:password@localhost"), + #{scheme := "foo", path := "/user@"} = + uri_string:parse("foo:/user@"), + #{scheme := "foo", userinfo := "user", host := "localhost"} = + uri_string:parse("foo://user@localhost"), + #{scheme := "foo", userinfo := "user:password", host := "localhost"} = + uri_string:parse("foo://user:password@localhost"). + +parse_pct_encoded_userinfo(_Config) -> + #{scheme := "user", path := "合@気道"} = + uri_string:parse("user:%E5%90%88@%E6%B0%97%E9%81%93"), + #{path := "合気道@"} = uri_string:parse("%E5%90%88%E6%B0%97%E9%81%93@"), + #{path := "/合気道@"} = uri_string:parse("/%E5%90%88%E6%B0%97%E9%81%93@"), + #{path := "合@気道"} = uri_string:parse("%E5%90%88@%E6%B0%97%E9%81%93"), + #{userinfo := "合", host := "気道"} = + uri_string:parse("//%E5%90%88@%E6%B0%97%E9%81%93"), + #{userinfo := "合:気", host := "道"} = + uri_string:parse("//%E5%90%88:%E6%B0%97@%E9%81%93"), + #{scheme := "foo", path := "/合気道@"} = + uri_string:parse("foo:/%E5%90%88%E6%B0%97%E9%81%93@"), + #{scheme := "foo", userinfo := "合", host := "気道"} = + uri_string:parse("foo://%E5%90%88@%E6%B0%97%E9%81%93"), + #{scheme := "foo", userinfo := "合:気", host := "道"} = + uri_string:parse("foo://%E5%90%88:%E6%B0%97@%E9%81%93"), + {error,invalid_uri,"@"} = uri_string:parse("//%E5%90%88@%E6%B0%97%E9%81%93@"), + {error,invalid_uri,":"} = uri_string:parse("foo://%E5%90%88@%E6%B0%97%E9%81%93@"). + + +parse_host(_Config) -> + #{host := "hostname"} = uri_string:parse("//hostname"), + #{host := "hostname",scheme := "foo"} = uri_string:parse("foo://hostname"), + #{host := "hostname",scheme := "foo", userinfo := "user"} = + uri_string:parse("foo://user@hostname"). + +parse_host_ipv4(_Config) -> + #{host := "127.0.0.1"} = uri_string:parse("//127.0.0.1"), + #{host := "2001:0db8:0000:0000:0000:0000:1428:07ab"} = + uri_string:parse("//[2001:0db8:0000:0000:0000:0000:1428:07ab]"), + #{host := "127.0.0.1", path := "/over/there"} = uri_string:parse("//127.0.0.1/over/there"), + #{host := "127.0.0.1", query := "name=ferret"} = uri_string:parse("//127.0.0.1?name=ferret"), + #{host := "127.0.0.1", fragment := "nose"} = uri_string:parse("//127.0.0.1#nose"), + {error,invalid_uri,"x"} = uri_string:parse("//127.0.0.x"), + {error,invalid_uri,"1227.0.0.1"} = uri_string:parse("//1227.0.0.1"). + +parse_host_ipv6(_Config) -> + #{host := "::127.0.0.1"} = uri_string:parse("//[::127.0.0.1]"), + #{host := "::127.0.0.1", path := "/over/there"} = uri_string:parse("//[::127.0.0.1]/over/there"), + #{host := "::127.0.0.1", query := "name=ferret"} = + uri_string:parse("//[::127.0.0.1]?name=ferret"), + #{host := "::127.0.0.1", fragment := "nose"} = uri_string:parse("//[::127.0.0.1]#nose"), + {error,invalid_uri,"x"} = uri_string:parse("//[::127.0.0.x]"), + {error,invalid_uri,"::1227.0.0.1"} = uri_string:parse("//[::1227.0.0.1]"), + {error,invalid_uri,"G"} = uri_string:parse("//[2001:0db8:0000:0000:0000:0000:1428:G7ab]"). + +parse_port(_Config) -> + #{path:= "/:8042"} = + uri_string:parse("/:8042"), + #{host:= "", port := 8042} = + uri_string:parse("//:8042"), + #{host := "example.com", port:= 8042} = + uri_string:parse("//example.com:8042"), + #{scheme := "foo", path := "/:8042"} = + uri_string:parse("foo:/:8042"), + #{scheme := "foo", host := "", port := 8042} = + uri_string:parse("foo://:8042"), + #{scheme := "foo", host := "example.com", port := 8042} = + uri_string:parse("foo://example.com:8042"). + +parse_path(_Config) -> + #{path := "over/there"} = uri_string:parse("over/there"), + #{path := "/over/there"} = uri_string:parse("/over/there"), + #{scheme := "foo", path := "/over/there"} = + uri_string:parse("foo:/over/there"), + #{scheme := "foo", host := "example.com", path := "/over/there"} = + uri_string:parse("foo://example.com/over/there"), + #{scheme := "foo", host := "example.com", path := "/over/there", port := 8042} = + uri_string:parse("foo://example.com:8042/over/there"). + +parse_query(_Config) -> + #{scheme := "foo", query := "name=ferret"} = + uri_string:parse("foo:?name=ferret"), + #{scheme := "foo", path:= "over/there", query := "name=ferret"} = + uri_string:parse("foo:over/there?name=ferret"), + #{scheme := "foo", path:= "/over/there", query := "name=ferret"} = + uri_string:parse("foo:/over/there?name=ferret"), + #{scheme := "foo", host := "example.com", query := "name=ferret"} = + uri_string:parse("foo://example.com?name=ferret"), + #{scheme := "foo", host := "example.com", path := "/", query := "name=ferret"} = + uri_string:parse("foo://example.com/?name=ferret"), + + #{path := "", query := "name=ferret"} = + uri_string:parse("?name=ferret"), + #{path := "over/there", query := "name=ferret"} = + uri_string:parse("over/there?name=ferret"), + #{path := "/", query := "name=ferret"} = + uri_string:parse("/?name=ferret"), + #{path := "/over/there", query := "name=ferret"} = + uri_string:parse("/over/there?name=ferret"), + #{host := "example.com", query := "name=ferret"} = + uri_string:parse("//example.com?name=ferret"), + #{host := "example.com", path := "/", query := "name=ferret"} = + uri_string:parse("//example.com/?name=ferret"). + +parse_pct_encoded_query(_Config) -> + #{scheme := "foo", host := "example.com", path := "/", + query := "name=合気道"} = + uri_string:parse("foo://example.com/?name=%E5%90%88%E6%B0%97%E9%81%93"), + #{host := "example.com", path := "/", query := "name=合気道"} = + uri_string:parse("//example.com/?name=%E5%90%88%E6%B0%97%E9%81%93"). + +parse_fragment(_Config) -> + #{scheme := "foo", fragment := "nose"} = + uri_string:parse("foo:#nose"), + #{scheme := "foo", path:= "over/there", fragment := "nose"} = + uri_string:parse("foo:over/there#nose"), + #{scheme := "foo", path:= "/over/there", fragment := "nose"} = + uri_string:parse("foo:/over/there#nose"), + #{scheme := "foo", host := "example.com", fragment := "nose"} = + uri_string:parse("foo://example.com#nose"), + #{scheme := "foo", host := "example.com", path := "/", fragment := "nose"} = + uri_string:parse("foo://example.com/#nose"), + #{scheme := "foo", host := "example.com", fragment := "nose"} = + uri_string:parse("foo://example.com#nose"), + + #{fragment := "nose"} = + uri_string:parse("#nose"), + #{path := "over/there", fragment := "nose"} = + uri_string:parse("over/there#nose"), + #{path := "/", fragment := "nose"} = + uri_string:parse("/#nose"), + #{path := "/over/there", fragment := "nose"} = + uri_string:parse("/over/there#nose"), + #{host := "example.com", fragment := "nose"} = + uri_string:parse("//example.com#nose"), + #{host := "example.com", path := "/", fragment := "nose"} = + uri_string:parse("//example.com/#nose"). + +parse_pct_encoded_fragment(_Config) -> + #{scheme := "foo", host := "example.com", fragment := "合気道"} = + uri_string:parse("foo://example.com#%E5%90%88%E6%B0%97%E9%81%93"), + #{host := "example.com", path := "/", fragment := "合気道"} = + uri_string:parse("//example.com/#%E5%90%88%E6%B0%97%E9%81%93"). + +parse_list(_Config) -> + #{scheme := "foo", path := "bar:nisse"} = uri_string:parse("foo:bar:nisse"), + #{scheme := "foo", host := "example.com", port := 8042, + path := "/over/there", query := "name=ferret", fragment := "nose"} = + uri_string:parse("foo://example.com:8042/over/there?name=ferret#nose"), + #{scheme := "foo", userinfo := "admin:admin", host := "example.com", port := 8042, + path := "/over/there", query := "name=ferret", fragment := "nose"} = + uri_string:parse("foo://admin:[email protected]:8042/over/there?name=ferret#nose"). + +parse_binary(_Config) -> + #{scheme := <<"foo">>, path := <<"bar:nisse">>} = uri_string:parse(<<"foo:bar:nisse">>), + #{scheme := <<"foo">>, host := <<"example.com">>, port := 8042, + path := <<"/over/there">>, query := <<"name=ferret">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo://example.com:8042/over/there?name=ferret#nose">>), + #{scheme := <<"foo">>, userinfo := <<"admin:admin">>, host := <<"example.com">>, port := 8042, + path := <<"/over/there">>, query := <<"name=ferret">>, fragment := <<"nose">>} = + uri_string:parse(<<"foo://admin:[email protected]:8042/over/there?name=ferret#nose">>). + + +parse_mixed(_Config) -> + #{scheme := "foo", path := "bar"} = + uri_string:parse(lists:append("fo",<<"o:bar">>)), + #{scheme := "foo", path := "bar"} = + uri_string:parse(lists:append("foo:b",<<"ar">>)), + #{scheme := "foo", path := "bar:bar"} = + uri_string:parse([[102],[111,111],<<":bar">>,58,98,97,114]). + +parse_relative(_Config) -> + #{path := "/path"} = + uri_string:parse(lists:append("/pa",<<"th">>)), + #{path := "foo"} = + uri_string:parse(lists:append("fo",<<"o">>)). + +parse_special(_Config) -> + #{host := [],query := []} = uri_string:parse("//?"), + #{fragment := [],host := []} = uri_string:parse("//#"), + #{host := [],query := [],scheme := "foo"} = uri_string:parse("foo://?"), + #{fragment := [],host := [],scheme := "foo"} = uri_string:parse("foo://#"), + #{host := <<>>, path := <<"/">>} = uri_string:parse(<<"///">>), + #{host := <<"hostname">>} = uri_string:parse(<<"//hostname">>), + #{host := <<>>, path := <<"/hostname">>} = uri_string:parse(<<"///hostname">>), + #{host := [],path := "/",query := []} = uri_string:parse("///?"), + #{fragment := [],host := [],path := "/"} = uri_string:parse("///#"), + #{host := "foo",query := []} = uri_string:parse("//foo?"), + #{fragment := [],host := "foo"} = uri_string:parse("//foo#"), + #{host := "foo",path := "/"} = uri_string:parse("//foo/"), + #{host := "foo",query := [],scheme := "http"} = uri_string:parse("http://foo?"), + #{fragment := [],host := "foo",scheme := "http"} = uri_string:parse("http://foo#"), + #{host := "foo",path := "/",scheme := "http"} = uri_string:parse("http://foo/"), + #{fragment := [],host := "host",port := 80,scheme := "http"} = uri_string:parse("http://host:80#"), + #{host := "host",port := 80,query := [],scheme := "http"} = uri_string:parse("http://host:80?"), + #{path := [],query := []} = uri_string:parse("?"), + #{path := [],query := "?"} = uri_string:parse("??"), + #{path := [],query := "??"} = uri_string:parse("???"). + +parse_special2(_Config) -> + #{host := [],path := "/",port := 1,scheme := "a"} = uri_string:parse("a://:1/"), + #{path := "/a/",scheme := "a"} = uri_string:parse("a:/a/"), + #{host := [],path := [],userinfo := []} = uri_string:parse("//@"), + #{host := [],path := [],scheme := "foo",userinfo := []} = uri_string:parse("foo://@"), + #{host := [],path := "/",userinfo := []} = uri_string:parse("//@/"), + #{host := [],path := "/",scheme := "foo",userinfo := []} = uri_string:parse("foo://@/"), + #{host := "localhost",path := "/",port := undefined} = uri_string:parse("//localhost:/"), + #{host := [],path := [],port := undefined} = uri_string:parse("//:"). + +parse_negative(_Config) -> + {error,invalid_uri,"å"} = uri_string:parse("å"), + {error,invalid_uri,"å"} = uri_string:parse("aå:/foo"), + {error,invalid_uri,":"} = uri_string:parse("foo://usär@host"), + {error,invalid_uri,"ö"} = uri_string:parse("//host/path?foö=bar"), + {error,invalid_uri,"ö"} = uri_string:parse("//host/path#foö"), + {error,invalid_uri,"127.256.0.1"} = uri_string:parse("//127.256.0.1"), + {error,invalid_uri,":::127.0.0.1"} = uri_string:parse("//[:::127.0.0.1]"), + {error,invalid_utf8,<<0,0,0,246>>} = uri_string:parse("//%00%00%00%F6"), + {error,invalid_uri,"A"} = uri_string:parse("//localhost:A8"). + + +%%------------------------------------------------------------------------- +%% Recompose tests +%%------------------------------------------------------------------------- +recompose_fragment(_Config) -> + <<?FRAGMENT_ENC>> = uri_string:recompose(#{fragment => <<?FRAGMENT/utf8>>, path => <<>>}), + ?FRAGMENT_ENC = uri_string:recompose(#{fragment => ?FRAGMENT, path => ""}). + +recompose_parse_fragment(_Config) -> + <<?FRAGMENT_ENC>> = uri_string:recompose(uri_string:parse(<<?FRAGMENT_ENC>>)), + ?FRAGMENT_ENC = uri_string:recompose(uri_string:parse(?FRAGMENT_ENC)). + +recompose_query(_Config) -> + <<?QUERY_ENC>> = + uri_string:recompose(#{query => <<?QUERY/utf8>>, path => <<>>}), + <<?QUERY_ENC?FRAGMENT_ENC>> = + uri_string:recompose(#{query => <<?QUERY/utf8>>, + fragment => <<?FRAGMENT/utf8>>, + path => <<>>}), + "?name=%C3%B6rn" = + uri_string:recompose(#{query => "name=örn", path => ""}), + "?name=%C3%B6rn#n%C3%A4sa" = + uri_string:recompose(#{query => "name=örn", + fragment => "näsa", + path => ""}). + +recompose_parse_query(_Config) -> + <<"?name=%C3%B6rn">> = uri_string:recompose(uri_string:parse(<<"?name=%C3%B6rn">>)), + <<"?name=%C3%B6rn#n%C3%A4sa">> = + uri_string:recompose(uri_string:parse(<<"?name=%C3%B6rn#n%C3%A4sa">>)), + "?name=%C3%B6rn" = uri_string:recompose(uri_string:parse("?name=%C3%B6rn")), + "?name=%C3%B6rn#n%C3%A4sa" = uri_string:recompose(uri_string:parse("?name=%C3%B6rn#n%C3%A4sa")). + +recompose_path(_Config) -> + <<"/d%C3%A4r">> = + uri_string:recompose(#{path => <<"/där"/utf8>>}), + <<"/d%C3%A4r#n%C3%A4sa">> = + uri_string:recompose(#{path => <<"/där"/utf8>>, + fragment => <<"näsa"/utf8>>}), + <<"/d%C3%A4r?name=%C3%B6rn">> = + uri_string:recompose(#{path => <<"/där"/utf8>>, + query => <<"name=örn"/utf8>>}), + <<"/d%C3%A4r?name=%C3%B6rn#n%C3%A4sa">> = + uri_string:recompose(#{path => <<"/där"/utf8>>, + query => <<"name=örn"/utf8>>, + fragment => <<"näsa"/utf8>>}), + + + "/d%C3%A4r" = + uri_string:recompose(#{path => "/där"}), + "/d%C3%A4r#n%C3%A4sa" = + uri_string:recompose(#{path => "/där", + fragment => "näsa"}), + "/d%C3%A4r?name=%C3%B6rn" = + uri_string:recompose(#{path => "/där", + query => "name=örn"}), + "/d%C3%A4r?name=%C3%B6rn#n%C3%A4sa" = + uri_string:recompose(#{path => "/där", + query => "name=örn", + fragment => "näsa"}). + + +recompose_parse_path(_Config) -> + <<"/d%C3%A4r">> = + uri_string:recompose(uri_string:parse(<<"/d%C3%A4r">>)), + <<"/d%C3%A4r#n%C3%A4sa">> = + uri_string:recompose(uri_string:parse(<<"/d%C3%A4r#n%C3%A4sa">>)), + <<"/d%C3%A4r?name=%C3%B6rn">> = + uri_string:recompose(uri_string:parse(<<"/d%C3%A4r?name=%C3%B6rn">>)), + + "/d%C3%A4r" = + uri_string:recompose(uri_string:parse("/d%C3%A4r")), + "/d%C3%A4r#n%C3%A4sa" = + uri_string:recompose(uri_string:parse("/d%C3%A4r#n%C3%A4sa")), + "/d%C3%A4r?name=%C3%B6rn" = + uri_string:recompose(uri_string:parse("/d%C3%A4r?name=%C3%B6rn")). + + +recompose_autogen(_Config) -> + Tests = generate_test_vectors(uri_combinations()), + lists:map(fun run_test_recompose/1, Tests). + +parse_recompose_autogen(_Config) -> + Tests = generate_test_vectors(uri_combinations()), + lists:map(fun run_test_parse_recompose/1, Tests). + +transcode_basic(_Config) -> + <<"foo%C3%B6bar"/utf8>> = + uri_string:transcode(<<"foo%00%00%00%F6bar"/utf32>>, [{in_encoding, utf32},{out_encoding, utf8}]), + "foo%C3%B6bar" = + uri_string:transcode("foo%00%00%00%F6bar", [{in_encoding, utf32},{out_encoding, utf8}]), + <<"foo%00%00%00%F6bar"/utf32>> = + uri_string:transcode(<<"foo%C3%B6bar"/utf8>>, [{in_encoding, utf8},{out_encoding, utf32}]), + "foo%00%00%00%F6bar" = + uri_string:transcode("foo%C3%B6bar", [{in_encoding, utf8},{out_encoding, utf32}]), + "foo%C3%B6bar" = + uri_string:transcode("foo%F6bar", [{in_encoding, latin1},{out_encoding, utf8}]). + +transcode_options(_Config) -> + <<"foo%C3%B6bar"/utf8>> = + uri_string:transcode(<<"foo%C3%B6bar"/utf8>>, []), + <<"foo%C3%B6bar"/utf8>> = + uri_string:transcode(<<"foo%00%00%00%F6bar"/utf32>>, [{in_encoding, utf32}]), + <<"foo%00%00%00%F6bar"/utf32>> = + uri_string:transcode(<<"foo%C3%B6bar"/utf8>>, [{out_encoding, utf32}]). + +transcode_mixed(_Config) -> + "foo%00%00%00%F6bar" = + uri_string:transcode(["foo",<<"%C3%B6"/utf8>>,<<"ba"/utf8>>,"r"], [{out_encoding, utf32}]), + "foo%00%00%00%F6bar" = + uri_string:transcode(["foo",<<"%C3%"/utf8>>,<<"B6ba"/utf8>>,"r"], [{out_encoding, utf32}]), + "foo%C3%B6bar" = + uri_string:transcode(["foo%00", <<"%00%0"/utf32>>,<<"0%F"/utf32>>,"6bar"], [{in_encoding, utf32},{out_encoding, utf8}]). + +transcode_negative(_Config) -> + {error,invalid_percent_encoding,"%BXbar"} = + uri_string:transcode(<<"foo%C3%BXbar"/utf8>>, [{in_encoding, utf8},{out_encoding, utf32}]), + {error,invalid_input,<<"ö">>} = + uri_string:transcode("foo%F6bar", [{in_encoding, utf8},{out_encoding, utf8}]). + +normalize(_Config) -> + "/a/g" = uri_string:normalize("/a/b/c/./../../g"), + <<"mid/6">> = uri_string:normalize(<<"mid/content=5/../6">>), + "http://localhost-%C3%B6rebro/a/g" = + uri_string:normalize("http://localhos%74-%c3%b6rebro:80/a/b/c/./../../g"), + <<"http://localhost-%C3%B6rebro/a/g">> = + uri_string:normalize(<<"http://localhos%74-%c3%b6rebro:80/a/b/c/./../../g">>), + <<"https://localhost/">> = + uri_string:normalize(<<"https://localhost:443">>), + <<"https://localhost:445/">> = + uri_string:normalize(<<"https://localhost:445">>), + <<"ftp://localhost">> = + uri_string:normalize(<<"ftp://localhost:21">>), + <<"ssh://localhost">> = + uri_string:normalize(<<"ssh://localhost:22">>), + <<"sftp://localhost">> = + uri_string:normalize(<<"sftp://localhost:22">>), + <<"tftp://localhost">> = + uri_string:normalize(<<"tftp://localhost:69">>). diff --git a/lib/stdlib/test/uri_string_property_test_SUITE.erl b/lib/stdlib/test/uri_string_property_test_SUITE.erl new file mode 100644 index 0000000000..ae2c61c7aa --- /dev/null +++ b/lib/stdlib/test/uri_string_property_test_SUITE.erl @@ -0,0 +1,39 @@ +%% +%% %CopyrightBegin% +%% +%% Copyright Ericsson AB 2008-2017. All Rights Reserved. +%% +%% Licensed under the Apache License, Version 2.0 (the "License"); +%% you may not use this file except in compliance with the License. +%% You may obtain a copy of the License at +%% +%% http://www.apache.org/licenses/LICENSE-2.0 +%% +%% Unless required by applicable law or agreed to in writing, software +%% distributed under the License is distributed on an "AS IS" BASIS, +%% WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. +%% See the License for the specific language governing permissions and +%% limitations under the License. +%% +%% %CopyrightEnd% +%% +-module(uri_string_property_test_SUITE). + +-include_lib("common_test/include/ct.hrl"). +-compile(export_all). + +all() -> [recompose]. + +init_per_suite(Config) -> + ct_property_test:init_per_suite(Config). + +end_per_suite(Config) -> + Config. + +%%%======================================================================== +%%% Test suites +%%%======================================================================== +recompose(Config) -> + ct_property_test:quickcheck( + uri_string_recompose:prop_recompose(), + Config). diff --git a/otp_versions.table b/otp_versions.table index 0cb8c9aa0b..157ca8f0ba 100644 --- a/otp_versions.table +++ b/otp_versions.table @@ -1,3 +1,4 @@ +OTP-20.1.5 : erts-9.1.4 inets-6.4.4 # asn1-5.0.3 common_test-1.15.2 compiler-7.1.3 cosEvent-2.2.1 cosEventDomain-1.2.1 cosFileTransfer-1.2.1 cosNotification-1.2.2 cosProperty-1.2.2 cosTime-1.2.2 cosTransactions-1.3.2 crypto-4.1 debugger-4.2.3 dialyzer-3.2.2 diameter-2.1.2 edoc-0.9.1 eldap-1.2.2 erl_docgen-0.7.1 erl_interface-3.10 et-1.6.1 eunit-2.3.4 hipe-3.16.1 ic-4.4.2 jinterface-1.8 kernel-5.4 megaco-3.18.2 mnesia-4.15.1 observer-2.5 odbc-2.12 orber-3.8.3 os_mon-2.4.3 otp_mibs-1.1.1 parsetools-2.1.5 public_key-1.5 reltool-0.7.5 runtime_tools-1.12.2 sasl-3.1 snmp-5.2.8 ssh-4.6.1 ssl-8.2.1 stdlib-3.4.2 syntax_tools-2.1.3 tools-2.11 wx-1.8.2 xmerl-1.3.15 : OTP-20.1.4 : inets-6.4.3 # asn1-5.0.3 common_test-1.15.2 compiler-7.1.3 cosEvent-2.2.1 cosEventDomain-1.2.1 cosFileTransfer-1.2.1 cosNotification-1.2.2 cosProperty-1.2.2 cosTime-1.2.2 cosTransactions-1.3.2 crypto-4.1 debugger-4.2.3 dialyzer-3.2.2 diameter-2.1.2 edoc-0.9.1 eldap-1.2.2 erl_docgen-0.7.1 erl_interface-3.10 erts-9.1.3 et-1.6.1 eunit-2.3.4 hipe-3.16.1 ic-4.4.2 jinterface-1.8 kernel-5.4 megaco-3.18.2 mnesia-4.15.1 observer-2.5 odbc-2.12 orber-3.8.3 os_mon-2.4.3 otp_mibs-1.1.1 parsetools-2.1.5 public_key-1.5 reltool-0.7.5 runtime_tools-1.12.2 sasl-3.1 snmp-5.2.8 ssh-4.6.1 ssl-8.2.1 stdlib-3.4.2 syntax_tools-2.1.3 tools-2.11 wx-1.8.2 xmerl-1.3.15 : OTP-20.1.3 : diameter-2.1.2 erts-9.1.3 snmp-5.2.8 # asn1-5.0.3 common_test-1.15.2 compiler-7.1.3 cosEvent-2.2.1 cosEventDomain-1.2.1 cosFileTransfer-1.2.1 cosNotification-1.2.2 cosProperty-1.2.2 cosTime-1.2.2 cosTransactions-1.3.2 crypto-4.1 debugger-4.2.3 dialyzer-3.2.2 edoc-0.9.1 eldap-1.2.2 erl_docgen-0.7.1 erl_interface-3.10 et-1.6.1 eunit-2.3.4 hipe-3.16.1 ic-4.4.2 inets-6.4.2 jinterface-1.8 kernel-5.4 megaco-3.18.2 mnesia-4.15.1 observer-2.5 odbc-2.12 orber-3.8.3 os_mon-2.4.3 otp_mibs-1.1.1 parsetools-2.1.5 public_key-1.5 reltool-0.7.5 runtime_tools-1.12.2 sasl-3.1 ssh-4.6.1 ssl-8.2.1 stdlib-3.4.2 syntax_tools-2.1.3 tools-2.11 wx-1.8.2 xmerl-1.3.15 : OTP-20.1.2 : diameter-2.1.1 erts-9.1.2 # asn1-5.0.3 common_test-1.15.2 compiler-7.1.3 cosEvent-2.2.1 cosEventDomain-1.2.1 cosFileTransfer-1.2.1 cosNotification-1.2.2 cosProperty-1.2.2 cosTime-1.2.2 cosTransactions-1.3.2 crypto-4.1 debugger-4.2.3 dialyzer-3.2.2 edoc-0.9.1 eldap-1.2.2 erl_docgen-0.7.1 erl_interface-3.10 et-1.6.1 eunit-2.3.4 hipe-3.16.1 ic-4.4.2 inets-6.4.2 jinterface-1.8 kernel-5.4 megaco-3.18.2 mnesia-4.15.1 observer-2.5 odbc-2.12 orber-3.8.3 os_mon-2.4.3 otp_mibs-1.1.1 parsetools-2.1.5 public_key-1.5 reltool-0.7.5 runtime_tools-1.12.2 sasl-3.1 snmp-5.2.7 ssh-4.6.1 ssl-8.2.1 stdlib-3.4.2 syntax_tools-2.1.3 tools-2.11 wx-1.8.2 xmerl-1.3.15 : |