summaryrefslogtreecommitdiff
path: root/Build/source/libs/luajit
diff options
context:
space:
mode:
Diffstat (limited to 'Build/source/libs/luajit')
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/dynasm/dasm_mips64.lua12
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/_lj_bcdef.h_226
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/_lj_ffdef.h_210
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/_lj_folddef.h_1068
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/_lj_libdef.h_393
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/_lj_recdef.h_263
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64.lua1216
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64be.lua12
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64.lua17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64el.lua17
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h4
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm64.h2022
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm64.h419
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/lj_load.c11
-rw-r--r--Build/source/libs/luajit/LuaJIT-src/src/vm_mips64.dasc5062
-rw-r--r--Build/source/libs/luajit/TLpatches/patch-046
-rw-r--r--Build/source/libs/luajit/TLpatches/patch-0913
17 files changed, 10967 insertions, 4 deletions
diff --git a/Build/source/libs/luajit/LuaJIT-src/dynasm/dasm_mips64.lua b/Build/source/libs/luajit/LuaJIT-src/dynasm/dasm_mips64.lua
new file mode 100644
index 00000000000..5636b23a6ec
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/dynasm/dasm_mips64.lua
@@ -0,0 +1,12 @@
+------------------------------------------------------------------------------
+-- DynASM MIPS64 module.
+--
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
+-- See dynasm.lua for full copyright notice.
+------------------------------------------------------------------------------
+-- This module just sets 64 bit mode for the combined MIPS/MIPS64 module.
+-- All the interesting stuff is there.
+------------------------------------------------------------------------------
+
+mips64 = true -- Using a global is an ugly, but effective solution.
+return require("dasm_mips")
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/_lj_bcdef.h_ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_bcdef.h_
new file mode 100644
index 00000000000..231ed3a4e44
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_bcdef.h_
@@ -0,0 +1,226 @@
+/* This is a generated file. DO NOT EDIT! */
+
+LJ_DATADEF const uint16_t lj_bc_ofs[] = {
+0,
+72,
+144,
+216,
+288,
+427,
+569,
+632,
+695,
+764,
+833,
+886,
+938,
+989,
+1040,
+1081,
+1122,
+1148,
+1180,
+1240,
+1314,
+1368,
+1422,
+1476,
+1530,
+1589,
+1643,
+1697,
+1751,
+1805,
+1841,
+1908,
+1975,
+2042,
+2109,
+2158,
+2230,
+2306,
+2342,
+2378,
+2408,
+2437,
+2462,
+2505,
+2541,
+2628,
+2710,
+2748,
+2782,
+2833,
+2897,
+3006,
+3099,
+3117,
+3135,
+3283,
+3407,
+3506,
+3679,
+3908,
+4032,
+4174,
+4220,
+4262,
+4266,
+4414,
+4482,
+4647,
+4838,
+4926,
+4930,
+5066,
+5158,
+5263,
+5360,
+5465,
+5485,
+5555,
+5622,
+5642,
+5686,
+5725,
+5745,
+5763,
+5810,
+5835,
+5855,
+5918,
+5972,
+5972,
+6097,
+6098,
+6177,
+7841,
+7908,
+8419,
+8522,
+8579,
+8710,
+7974,
+8136,
+8228,
+8280,
+8311,
+8768,
+8809,
+9417,
+8864,
+9167,
+9469,
+9596,
+9620,
+9647,
+9711,
+9744,
+9778,
+9809,
+9840,
+9873,
+9914,
+9957,
+9990,
+10030,
+10070,
+10245,
+10393,
+10110,
+10110,
+9678,
+10149,
+10549,
+10492,
+10196,
+10603,
+10662,
+11596,
+11994,
+11941,
+12063,
+12142,
+12224,
+12306,
+12388,
+11650,
+11747,
+11844,
+10721,
+10752,
+10799,
+10921,
+11090,
+11217,
+11327,
+11442,
+11557
+};
+
+LJ_DATADEF const uint16_t lj_bc_mode[] = {
+BCDEF(BCMODE)
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF,
+BCMODE_FF
+};
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/_lj_ffdef.h_ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_ffdef.h_
new file mode 100644
index 00000000000..29dd4d3edce
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_ffdef.h_
@@ -0,0 +1,210 @@
+/* This is a generated file. DO NOT EDIT! */
+
+FFDEF(assert)
+FFDEF(type)
+FFDEF(next)
+FFDEF(pairs)
+FFDEF(ipairs_aux)
+FFDEF(ipairs)
+FFDEF(getmetatable)
+FFDEF(setmetatable)
+FFDEF(getfenv)
+FFDEF(setfenv)
+FFDEF(rawget)
+FFDEF(rawset)
+FFDEF(rawequal)
+FFDEF(unpack)
+FFDEF(select)
+FFDEF(tonumber)
+FFDEF(tostring)
+FFDEF(error)
+FFDEF(pcall)
+FFDEF(xpcall)
+FFDEF(loadfile)
+FFDEF(load)
+FFDEF(loadstring)
+FFDEF(dofile)
+FFDEF(gcinfo)
+FFDEF(collectgarbage)
+FFDEF(newproxy)
+FFDEF(print)
+FFDEF(coroutine_status)
+FFDEF(coroutine_running)
+FFDEF(coroutine_create)
+FFDEF(coroutine_yield)
+FFDEF(coroutine_resume)
+FFDEF(coroutine_wrap_aux)
+FFDEF(coroutine_wrap)
+FFDEF(math_abs)
+FFDEF(math_floor)
+FFDEF(math_ceil)
+FFDEF(math_sqrt)
+FFDEF(math_log10)
+FFDEF(math_exp)
+FFDEF(math_sin)
+FFDEF(math_cos)
+FFDEF(math_tan)
+FFDEF(math_asin)
+FFDEF(math_acos)
+FFDEF(math_atan)
+FFDEF(math_sinh)
+FFDEF(math_cosh)
+FFDEF(math_tanh)
+FFDEF(math_frexp)
+FFDEF(math_modf)
+FFDEF(math_deg)
+FFDEF(math_rad)
+FFDEF(math_log)
+FFDEF(math_atan2)
+FFDEF(math_pow)
+FFDEF(math_fmod)
+FFDEF(math_ldexp)
+FFDEF(math_min)
+FFDEF(math_max)
+FFDEF(math_random)
+FFDEF(math_randomseed)
+FFDEF(bit_tobit)
+FFDEF(bit_bnot)
+FFDEF(bit_bswap)
+FFDEF(bit_lshift)
+FFDEF(bit_rshift)
+FFDEF(bit_arshift)
+FFDEF(bit_rol)
+FFDEF(bit_ror)
+FFDEF(bit_band)
+FFDEF(bit_bor)
+FFDEF(bit_bxor)
+FFDEF(bit_tohex)
+FFDEF(string_len)
+FFDEF(string_byte)
+FFDEF(string_char)
+FFDEF(string_sub)
+FFDEF(string_rep)
+FFDEF(string_reverse)
+FFDEF(string_lower)
+FFDEF(string_upper)
+FFDEF(string_dump)
+FFDEF(string_find)
+FFDEF(string_match)
+FFDEF(string_gmatch_aux)
+FFDEF(string_gmatch)
+FFDEF(string_gsub)
+FFDEF(string_format)
+FFDEF(table_foreachi)
+FFDEF(table_foreach)
+FFDEF(table_getn)
+FFDEF(table_maxn)
+FFDEF(table_insert)
+FFDEF(table_remove)
+FFDEF(table_concat)
+FFDEF(table_sort)
+FFDEF(io_method_close)
+FFDEF(io_method_read)
+FFDEF(io_method_write)
+FFDEF(io_method_flush)
+FFDEF(io_method_seek)
+FFDEF(io_method_setvbuf)
+FFDEF(io_method_lines)
+FFDEF(io_method___gc)
+FFDEF(io_method___tostring)
+FFDEF(io_open)
+FFDEF(io_popen)
+FFDEF(io_tmpfile)
+FFDEF(io_close)
+FFDEF(io_read)
+FFDEF(io_write)
+FFDEF(io_flush)
+FFDEF(io_input)
+FFDEF(io_output)
+FFDEF(io_lines)
+FFDEF(io_type)
+FFDEF(os_execute)
+FFDEF(os_remove)
+FFDEF(os_rename)
+FFDEF(os_tmpname)
+FFDEF(os_getenv)
+FFDEF(os_exit)
+FFDEF(os_clock)
+FFDEF(os_date)
+FFDEF(os_time)
+FFDEF(os_difftime)
+FFDEF(os_setlocale)
+FFDEF(debug_getregistry)
+FFDEF(debug_getmetatable)
+FFDEF(debug_setmetatable)
+FFDEF(debug_getfenv)
+FFDEF(debug_setfenv)
+FFDEF(debug_getinfo)
+FFDEF(debug_getlocal)
+FFDEF(debug_setlocal)
+FFDEF(debug_getupvalue)
+FFDEF(debug_setupvalue)
+FFDEF(debug_upvalueid)
+FFDEF(debug_upvaluejoin)
+FFDEF(debug_sethook)
+FFDEF(debug_gethook)
+FFDEF(debug_debug)
+FFDEF(debug_traceback)
+FFDEF(jit_on)
+FFDEF(jit_off)
+FFDEF(jit_flush)
+FFDEF(jit_status)
+FFDEF(jit_attach)
+FFDEF(jit_util_funcinfo)
+FFDEF(jit_util_funcbc)
+FFDEF(jit_util_funck)
+FFDEF(jit_util_funcuvname)
+FFDEF(jit_util_traceinfo)
+FFDEF(jit_util_traceir)
+FFDEF(jit_util_tracek)
+FFDEF(jit_util_tracesnap)
+FFDEF(jit_util_tracemc)
+FFDEF(jit_util_traceexitstub)
+FFDEF(jit_util_ircalladdr)
+FFDEF(jit_opt_start)
+FFDEF(ffi_meta___index)
+FFDEF(ffi_meta___newindex)
+FFDEF(ffi_meta___eq)
+FFDEF(ffi_meta___len)
+FFDEF(ffi_meta___lt)
+FFDEF(ffi_meta___le)
+FFDEF(ffi_meta___concat)
+FFDEF(ffi_meta___call)
+FFDEF(ffi_meta___add)
+FFDEF(ffi_meta___sub)
+FFDEF(ffi_meta___mul)
+FFDEF(ffi_meta___div)
+FFDEF(ffi_meta___mod)
+FFDEF(ffi_meta___pow)
+FFDEF(ffi_meta___unm)
+FFDEF(ffi_meta___tostring)
+FFDEF(ffi_meta___pairs)
+FFDEF(ffi_meta___ipairs)
+FFDEF(ffi_clib___index)
+FFDEF(ffi_clib___newindex)
+FFDEF(ffi_clib___gc)
+FFDEF(ffi_callback_free)
+FFDEF(ffi_callback_set)
+FFDEF(ffi_cdef)
+FFDEF(ffi_new)
+FFDEF(ffi_cast)
+FFDEF(ffi_typeof)
+FFDEF(ffi_istype)
+FFDEF(ffi_sizeof)
+FFDEF(ffi_alignof)
+FFDEF(ffi_offsetof)
+FFDEF(ffi_errno)
+FFDEF(ffi_string)
+FFDEF(ffi_copy)
+FFDEF(ffi_fill)
+FFDEF(ffi_abi)
+FFDEF(ffi_metatype)
+FFDEF(ffi_gc)
+FFDEF(ffi_load)
+
+#undef FFDEF
+
+#ifndef FF_NUM_ASMFUNC
+#define FF_NUM_ASMFUNC 62
+#endif
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/_lj_folddef.h_ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_folddef.h_
new file mode 100644
index 00000000000..c4ec4efd54e
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_folddef.h_
@@ -0,0 +1,1068 @@
+/* This is a generated file. DO NOT EDIT! */
+
+static const FoldFunc fold_func[] = {
+ fold_kfold_numarith,
+ fold_kfold_ldexp,
+ fold_kfold_fpmath,
+ fold_kfold_numpow,
+ fold_kfold_numcomp,
+ fold_kfold_intarith,
+ fold_kfold_intovarith,
+ fold_kfold_bnot,
+ fold_kfold_bswap,
+ fold_kfold_intcomp,
+ fold_kfold_intcomp0,
+ fold_kfold_int64arith,
+ fold_kfold_int64arith2,
+ fold_kfold_int64shift,
+ fold_kfold_bnot64,
+ fold_kfold_bswap64,
+ fold_kfold_int64comp,
+ fold_kfold_int64comp0,
+ fold_kfold_snew_kptr,
+ fold_kfold_snew_empty,
+ fold_kfold_strref,
+ fold_kfold_strref_snew,
+ fold_kfold_strcmp,
+ fold_kfold_add_kgc,
+ fold_kfold_add_kptr,
+ fold_kfold_add_kright,
+ fold_kfold_tobit,
+ fold_kfold_conv_kint_num,
+ fold_kfold_conv_kintu32_num,
+ fold_kfold_conv_kint_ext,
+ fold_kfold_conv_kint_i64,
+ fold_kfold_conv_kint64_num_i64,
+ fold_kfold_conv_kint64_num_u64,
+ fold_kfold_conv_kint64_int_i64,
+ fold_kfold_conv_knum_int_num,
+ fold_kfold_conv_knum_u32_num,
+ fold_kfold_conv_knum_i64_num,
+ fold_kfold_conv_knum_u64_num,
+ fold_kfold_tostr_knum,
+ fold_kfold_tostr_kint,
+ fold_kfold_strto,
+ lj_opt_cse,
+ fold_kfold_kref,
+ fold_shortcut_round,
+ fold_shortcut_left,
+ fold_shortcut_dropleft,
+ fold_shortcut_leftleft,
+ fold_simplify_numadd_negx,
+ fold_simplify_numadd_xneg,
+ fold_simplify_numsub_k,
+ fold_simplify_numsub_negk,
+ fold_simplify_numsub_xneg,
+ fold_simplify_nummuldiv_k,
+ fold_simplify_nummuldiv_negk,
+ fold_simplify_nummuldiv_negneg,
+ fold_simplify_numpow_xk,
+ fold_simplify_numpow_kx,
+ fold_shortcut_conv_num_int,
+ fold_simplify_conv_int_num,
+ fold_simplify_conv_i64_num,
+ fold_simplify_conv_int_i64,
+ fold_simplify_conv_flt_num,
+ fold_simplify_tobit_conv,
+ fold_simplify_floor_conv,
+ fold_simplify_conv_sext,
+ fold_simplify_conv_narrow,
+ fold_cse_conv,
+ fold_narrow_convert,
+ fold_simplify_intadd_k,
+ fold_simplify_intmul_k,
+ fold_simplify_intsub_k,
+ fold_simplify_intsub_kleft,
+ fold_simplify_intadd_k64,
+ fold_simplify_intsub_k64,
+ fold_simplify_intmul_k32,
+ fold_simplify_intmul_k64,
+ fold_simplify_intmod_k,
+ fold_simplify_intmod_kleft,
+ fold_simplify_intsub,
+ fold_simplify_intsubadd_leftcancel,
+ fold_simplify_intsubsub_leftcancel,
+ fold_simplify_intsubsub_rightcancel,
+ fold_simplify_intsubadd_rightcancel,
+ fold_simplify_intsubaddadd_cancel,
+ fold_simplify_band_k,
+ fold_simplify_bor_k,
+ fold_simplify_bxor_k,
+ fold_simplify_shift_ik,
+ fold_simplify_shift_andk,
+ fold_simplify_shift1_ki,
+ fold_simplify_shift2_ki,
+ fold_simplify_shiftk_andk,
+ fold_simplify_andk_shiftk,
+ fold_reassoc_intarith_k,
+ fold_reassoc_intarith_k64,
+ fold_reassoc_dup,
+ fold_reassoc_bxor,
+ fold_reassoc_shift,
+ fold_reassoc_minmax_k,
+ fold_reassoc_minmax_left,
+ fold_reassoc_minmax_right,
+ fold_abc_fwd,
+ fold_abc_k,
+ fold_abc_invar,
+ fold_comm_swap,
+ fold_comm_equal,
+ fold_comm_comp,
+ fold_comm_dup,
+ fold_comm_bxor,
+ fold_merge_eqne_snew_kgc,
+ lj_opt_fwd_aload,
+ fold_kfold_hload_kkptr,
+ lj_opt_fwd_hload,
+ lj_opt_fwd_uload,
+ lj_opt_fwd_tab_len,
+ fold_cse_uref,
+ lj_opt_fwd_hrefk,
+ fold_fwd_href_tnew,
+ fold_fwd_href_tdup,
+ fold_fload_tab_tnew_asize,
+ fold_fload_tab_tnew_hmask,
+ fold_fload_tab_tdup_asize,
+ fold_fload_tab_tdup_hmask,
+ fold_fload_tab_ah,
+ fold_fload_str_len_kgc,
+ fold_fload_str_len_snew,
+ fold_fload_cdata_typeid_kgc,
+ fold_fload_cdata_int64_kgc,
+ fold_fload_cdata_typeid_cnew,
+ fold_fload_cdata_ptr_int64_cnew,
+ lj_opt_cse,
+ lj_opt_fwd_fload,
+ fold_fwd_sload,
+ fold_xload_kptr,
+ lj_opt_fwd_xload,
+ fold_barrier_tab,
+ fold_barrier_tnew_tdup,
+ lj_opt_dse_ahstore,
+ lj_opt_dse_ustore,
+ lj_opt_dse_fstore,
+ lj_opt_dse_xstore,
+ lj_ir_emit
+};
+
+static const uint32_t fold_hash[916] = {
+0xffffffff,
+0xffffffff,
+0x5b4c8016,
+0x0d4e7016,
+0xffffffff,
+0x1000701c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x29110c1a,
+0xffffffff,
+0xffffffff,
+0x5b488016,
+0x0d4a7016,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x7b87fc07,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x0d467016,
+0xffffffff,
+0x5a4c73ff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5153fc29,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5d408016,
+0xffffffff,
+0x594873ff,
+0x8187440f,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8287fc0f,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6715ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a11fc1a,
+0xffffffff,
+0x1daa5a70,
+0xffffffff,
+0xffffffff,
+0x0a0bfc16,
+0x5c408c16,
+0x6911ffff,
+0x8db7ffff,
+0xffffffff,
+0xffffffff,
+0x1caa59d4,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6a0dffff,
+0x2b68d002,
+0xffffffff,
+0x3cab5695,
+0xffffffff,
+0x41aaa675,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x27ae5800,
+0xffffffff,
+0x6a09ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x7f865c0f,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6a05ffff,
+0x42abffff,
+0x5e44881c,
+0x5d50a016,
+0x066c5816,
+0x00646c1b,
+0x75753bff,
+0x1951fc18,
+0x6264c81b,
+0x1850641c,
+0xffffffff,
+0x6a01ffff,
+0x87a7ffff,
+0x4953fc1c,
+0x8da80000,
+0x4f52a3ff,
+0x00606c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5d428416,
+0x88a53800,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x05645816,
+0xffffffff,
+0x005c6c1b,
+0x20aa71d6,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x1399fc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x157f33ff,
+0xffffffff,
+0xffffffff,
+0x584dfc20,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8d9bffff,
+0xffffffff,
+0x055c5816,
+0xffffffff,
+0x00546c1b,
+0xffffffff,
+0xffffffff,
+0x5849fc20,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8c97ffff,
+0x5543fc1c,
+0x05585816,
+0xffffffff,
+0x00506c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8a93ffff,
+0x26ae6c00,
+0x05545816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x76753c17,
+0x41aaa695,
+0xffffffff,
+0x898fffff,
+0xffffffff,
+0x05505816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x858867ff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x848bffff,
+0xffffffff,
+0x054c5816,
+0x79873c06,
+0x47525bff,
+0xffffffff,
+0x3f695401,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8387ffff,
+0xffffffff,
+0x05485816,
+0xffffffff,
+0x5a4e5bff,
+0xffffffff,
+0xffffffff,
+0x6264c816,
+0x43aaa26e,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x05445816,
+0xffffffff,
+0x5a4a5bff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3455fc1b,
+0x0c5a701c,
+0x6366cbff,
+0x0e3c7000,
+0xffffffff,
+0x05405816,
+0xffffffff,
+0x59465bff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x41aaa276,
+0x0c56701c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x21aa7275,
+0x0b52701c,
+0x61489016,
+0x6465fc33,
+0x8d77ffff,
+0xffffffff,
+0x7b87fc05,
+0xffffffff,
+0xffffffff,
+0x2a126bff,
+0x385a6fff,
+0xffffffff,
+0x446dfc16,
+0xffffffff,
+0x7473ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x7d873000,
+0xffffffff,
+0x5c409016,
+0x686fffff,
+0x8187440d,
+0xffffffff,
+0xffffffff,
+0x3554b81b,
+0x8287fc0d,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x686bffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8d9ffc00,
+0x737a5fff,
+0x41aaaa75,
+0xffffffff,
+0xffffffff,
+0x5e40801c,
+0x0b42701c,
+0x6b67ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2b68d000,
+0xffffffff,
+0xffffffff,
+0x6d133017,
+0xffffffff,
+0xffffffff,
+0x4c59fc16,
+0xffffffff,
+0xffffffff,
+0x110bfc1c,
+0x3aab566e,
+0xffffffff,
+0x5052a7ff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6515fc28,
+0x4a55fc16,
+0x7f865c0d,
+0x88a53c00,
+0x41aaa296,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x4451fc16,
+0xffffffff,
+0x60448bff,
+0x21aa7295,
+0xffffffff,
+0x3cab5676,
+0x04106c1b,
+0xffffffff,
+0x78873807,
+0xffffffff,
+0xffffffff,
+0x574dfc16,
+0xffffffff,
+0x4e53ffff,
+0xffffffff,
+0x09145816,
+0xffffffff,
+0x040c6c1b,
+0x8287fc00,
+0x5e50a01c,
+0x6467fc32,
+0xffffffff,
+0x5749fc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a105816,
+0x2e3e7c00,
+0x04086c1b,
+0x7083fc00,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5645fc16,
+0xffffffff,
+0x22aa6e6e,
+0x5e42841c,
+0x614e9c16,
+0x090c5816,
+0x04046c1b,
+0x1eaa5ab3,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5441fc16,
+0x41aaaa95,
+0xffffffff,
+0x5352a028,
+0x09085816,
+0x17505c16,
+0x04006c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6b43ffff,
+0xffffffff,
+0x09045816,
+0xffffffff,
+0x43aaa2ae,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x083e5800,
+0x7c865c00,
+0xffffffff,
+0x76753c15,
+0x3051fc2e,
+0x09005816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3aab568e,
+0xffffffff,
+0x43aaa66e,
+0xffffffff,
+0x1daa5a71,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a12701c,
+0x5f66cfff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3cab5696,
+0xffffffff,
+0x100e701c,
+0x41aaa676,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a125c17,
+0x3654b82e,
+0x100a701c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x1006701c,
+0xffffffff,
+0x1951fc19,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x23aa6e8e,
+0xffffffff,
+0x5b4e8016,
+0xffffffff,
+0x1eaa5ad3,
+0x1002701c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x29130c1a,
+0xffffffff,
+0xffffffff,
+0x0d4c7016,
+0xffffffff,
+0x475273ff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5b468016,
+0x0d487016,
+0x5a4e73ff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5d54a816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5a4a73ff,
+0x6615fc16,
+0x3bab56ae,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x594673ff,
+0xffffffff,
+0x61468c16,
+0x8d17ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a13fc1a,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6913ffff,
+0x40abfeb3,
+0x8db9ffff,
+0x41aaa696,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6a0fffff,
+0x8db5ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x7a873c07,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6a0bffff,
+0x3f695402,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x24aa6eae,
+0xffffffff,
+0xffffffff,
+0x6a07ffff,
+0xffffffff,
+0xffffffff,
+0x066e5816,
+0xffffffff,
+0x00666c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6a03ffff,
+0xffffffff,
+0x4b55fc1c,
+0x066a5816,
+0xffffffff,
+0x00626c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x4851fc1c,
+0x05665816,
+0x18506016,
+0x005e6c1b,
+0x12986416,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8da1ffff,
+0xffffffff,
+0x3bab56ce,
+0xffffffff,
+0x43aaa6ae,
+0xffffffff,
+0xffffffff,
+0x584ffc20,
+0x7b87fc06,
+0xffffffff,
+0x5f4287ff,
+0x8d9dffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x00566c1b,
+0xffffffff,
+0xffffffff,
+0x584bfc20,
+0x5253fc28,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5645fc1c,
+0xffffffff,
+0x40abfed3,
+0x00526c1b,
+0x8187440e,
+0xffffffff,
+0x5847fc20,
+0x8287fc0e,
+0xffffffff,
+0xffffffff,
+0x8b95ffff,
+0x2e3c7800,
+0x5441fc1c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x17505c1c,
+0xffffffff,
+0xffffffff,
+0x41aaaa76,
+0xffffffff,
+0x614c9816,
+0x8991ffff,
+0x1daa5a6f,
+0x05525816,
+0x4d585bff,
+0xffffffff,
+0x8087400c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x1baa59d3,
+0x828dffff,
+0x25aa6ece,
+0x054e5816,
+0x76753c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2b68d001,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x8689ffff,
+0xffffffff,
+0x054a5816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x43aca01b,
+0x05465816,
+0x7f865c0e,
+0x5a4c5bff,
+0x39ab55d3,
+0x01626c16,
+0x02686fff,
+0x3457fc1b,
+0xffffffff,
+0xffffffff,
+0x0f3e7000,
+0x3dab55ae,
+0x05425816,
+0x1951fc17,
+0x59485bff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3153fc1b,
+0x0c58701c,
+0x5f64cbff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x035a6c16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x0b54701c,
+0xffffffff,
+0x8779ffff,
+0x1faa71d5,
+0xffffffff,
+0xffffffff,
+0x2d5eb81b,
+0x72b5fc08,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x0b50701c,
+0x456ffc16,
+0x7b75ffff,
+0xffffffff,
+0xffffffff,
+0x147e5c16,
+0xffffffff,
+0xffffffff,
+0x2a106bff,
+0xffffffff,
+0x1eaa5ab4,
+0x446bfc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x41aaaa96,
+0xffffffff,
+0x3556b81b,
+0x87a5fc00,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x4e6dffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3252b81b,
+0xffffffff,
+0x5e54a81c,
+0xffffffff,
+0xffffffff,
+0x0b44701c,
+0x28b05c00,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x73785fff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x0b40701c,
+0xffffffff,
+0x6b65ffff,
+0xffffffff,
+0xffffffff,
+0x1daa5a72,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6266cc1b,
+0xffffffff,
+0x375bfc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x3f695400,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6d113017,
+0x3ead541b,
+0xffffffff,
+0x5d448816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x18506416,
+0xffffffff,
+0xffffffff,
+0x16b37400,
+0xffffffff,
+0x4653fc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x04126c1b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x574ffc16,
+0xffffffff,
+0x6855ffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x040e6c1b,
+0x41aaa275,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x574bfc16,
+0x6f826400,
+0x6851ffff,
+0x1eaa5ad4,
+0x2a125816,
+0xffffffff,
+0x040a6c1b,
+0x7185fc00,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x5747fc16,
+0x7b87fc04,
+0xffffffff,
+0xffffffff,
+0x090e5816,
+0xffffffff,
+0x04066c1b,
+0x6e81fc00,
+0x1aac6c1b,
+0x1850601c,
+0x2e5cbbff,
+0x5543fc16,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x090a5816,
+0xffffffff,
+0x04026c1b,
+0xffffffff,
+0xffffffff,
+0x8087440c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6c45ffff,
+0x8287fc0c,
+0x09065816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x6b41ffff,
+0x3353fc2e,
+0x09025816,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2f50bbff,
+0x073c5800,
+0x6266cc16,
+0x5f4083ff,
+0xffffffff,
+0xffffffff,
+0x43aca41b,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2a10701c,
+0x6364cfff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x7e865c0c,
+0xffffffff,
+0xffffffff,
+0x3656b82e,
+0x41aaa295,
+0x100c701c,
+0x614a9416,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x2c5ebc1b,
+0xffffffff,
+0x2a105c17,
+0xffffffff,
+0x1008701c,
+0x3cab5675,
+0xffffffff,
+0xffffffff,
+0x77873806,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0x1004701c,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff,
+0xffffffff
+};
+
+#define fold_hashkey(k) (lj_rol(lj_rol((k),17)-(k),16)%915)
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/_lj_libdef.h_ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_libdef.h_
new file mode 100644
index 00000000000..5693a581391
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_libdef.h_
@@ -0,0 +1,393 @@
+/* This is a generated file. DO NOT EDIT! */
+
+#ifdef LJLIB_MODULE_base
+#undef LJLIB_MODULE_base
+static const lua_CFunction lj_lib_cf_base[] = {
+ lj_ffh_assert,
+ lj_ffh_next,
+ lj_ffh_pairs,
+ lj_ffh_ipairs_aux,
+ lj_ffh_ipairs,
+ lj_ffh_setmetatable,
+ lj_cf_getfenv,
+ lj_cf_setfenv,
+ lj_ffh_rawget,
+ lj_cf_rawset,
+ lj_cf_rawequal,
+ lj_cf_unpack,
+ lj_cf_select,
+ lj_ffh_tonumber,
+ lj_ffh_tostring,
+ lj_cf_error,
+ lj_ffh_pcall,
+ lj_cf_loadfile,
+ lj_cf_load,
+ lj_cf_loadstring,
+ lj_cf_dofile,
+ lj_cf_gcinfo,
+ lj_cf_collectgarbage,
+ lj_cf_newproxy,
+ lj_cf_print
+};
+static const uint8_t lj_lib_init_base[] = {
+2,0,28,70,97,115,115,101,114,116,195,110,105,108,199,98,111,111,108,101,97,
+110,252,1,200,117,115,101,114,100,97,116,97,198,115,116,114,105,110,103,197,
+117,112,118,97,108,198,116,104,114,101,97,100,197,112,114,111,116,111,200,102,
+117,110,99,116,105,111,110,197,116,114,97,99,101,197,99,100,97,116,97,197,116,
+97,98,108,101,252,9,198,110,117,109,98,101,114,132,116,121,112,101,68,110,101,
+120,116,253,69,112,97,105,114,115,64,253,70,105,112,97,105,114,115,140,103,
+101,116,109,101,116,97,116,97,98,108,101,76,115,101,116,109,101,116,97,116,
+97,98,108,101,7,103,101,116,102,101,110,118,7,115,101,116,102,101,110,118,70,
+114,97,119,103,101,116,6,114,97,119,115,101,116,8,114,97,119,101,113,117,97,
+108,6,117,110,112,97,99,107,6,115,101,108,101,99,116,72,116,111,110,117,109,
+98,101,114,195,110,105,108,197,102,97,108,115,101,196,116,114,117,101,72,116,
+111,115,116,114,105,110,103,5,101,114,114,111,114,69,112,99,97,108,108,134,
+120,112,99,97,108,108,8,108,111,97,100,102,105,108,101,4,108,111,97,100,10,
+108,111,97,100,115,116,114,105,110,103,6,100,111,102,105,108,101,6,103,99,105,
+110,102,111,14,99,111,108,108,101,99,116,103,97,114,98,97,103,101,252,2,8,110,
+101,119,112,114,111,120,121,200,116,111,115,116,114,105,110,103,5,112,114,105,
+110,116,252,3,200,95,86,69,82,83,73,79,78,250,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_coroutine
+#undef LJLIB_MODULE_coroutine
+static const lua_CFunction lj_lib_cf_coroutine[] = {
+ lj_cf_coroutine_status,
+ lj_cf_coroutine_running,
+ lj_cf_coroutine_create,
+ lj_ffh_coroutine_yield,
+ lj_ffh_coroutine_resume,
+ lj_cf_coroutine_wrap
+};
+static const uint8_t lj_lib_init_coroutine[] = {
+30,13,6,6,115,116,97,116,117,115,7,114,117,110,110,105,110,103,6,99,114,101,
+97,116,101,69,121,105,101,108,100,70,114,101,115,117,109,101,254,4,119,114,
+97,112,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_math
+#undef LJLIB_MODULE_math
+static const lua_CFunction lj_lib_cf_math[] = {
+ lj_ffh_math_abs,
+ lj_ffh_math_sqrt,
+ lj_ffh_math_log,
+ lj_ffh_math_atan2,
+ lj_ffh_math_ldexp,
+ lj_ffh_math_min,
+ lj_cf_math_random,
+ lj_cf_math_randomseed
+};
+static const uint8_t lj_lib_init_math[] = {
+37,16,30,67,97,98,115,133,102,108,111,111,114,132,99,101,105,108,68,115,113,
+114,116,133,108,111,103,49,48,131,101,120,112,131,115,105,110,131,99,111,115,
+131,116,97,110,132,97,115,105,110,132,97,99,111,115,132,97,116,97,110,132,115,
+105,110,104,132,99,111,115,104,132,116,97,110,104,133,102,114,101,120,112,132,
+109,111,100,102,251,248,193,99,26,220,165,76,64,131,100,101,103,251,57,157,
+82,162,70,223,145,63,131,114,97,100,67,108,111,103,69,97,116,97,110,50,131,
+112,111,119,132,102,109,111,100,69,108,100,101,120,112,67,109,105,110,131,109,
+97,120,251,24,45,68,84,251,33,9,64,194,112,105,250,251,0,0,0,0,0,0,240,127,
+196,104,117,103,101,250,252,2,6,114,97,110,100,111,109,252,2,10,114,97,110,
+100,111,109,115,101,101,100,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_bit
+#undef LJLIB_MODULE_bit
+static const lua_CFunction lj_lib_cf_bit[] = {
+ lj_ffh_bit_tobit,
+ lj_ffh_bit_lshift,
+ lj_ffh_bit_band,
+ lj_cf_bit_tohex
+};
+static const uint8_t lj_lib_init_bit[] = {
+65,42,12,69,116,111,98,105,116,132,98,110,111,116,133,98,115,119,97,112,70,
+108,115,104,105,102,116,134,114,115,104,105,102,116,135,97,114,115,104,105,
+102,116,131,114,111,108,131,114,111,114,68,98,97,110,100,131,98,111,114,132,
+98,120,111,114,5,116,111,104,101,120,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_string
+#undef LJLIB_MODULE_string
+static const lua_CFunction lj_lib_cf_string[] = {
+ lj_ffh_string_len,
+ lj_ffh_string_byte,
+ lj_ffh_string_char,
+ lj_ffh_string_sub,
+ lj_ffh_string_rep,
+ lj_ffh_string_reverse,
+ lj_cf_string_dump,
+ lj_cf_string_find,
+ lj_cf_string_match,
+ lj_cf_string_gmatch,
+ lj_cf_string_gsub,
+ lj_cf_string_format
+};
+static const uint8_t lj_lib_init_string[] = {
+77,53,14,67,108,101,110,68,98,121,116,101,68,99,104,97,114,67,115,117,98,67,
+114,101,112,71,114,101,118,101,114,115,101,133,108,111,119,101,114,133,117,
+112,112,101,114,4,100,117,109,112,4,102,105,110,100,5,109,97,116,99,104,254,
+6,103,109,97,116,99,104,4,103,115,117,98,6,102,111,114,109,97,116,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_table
+#undef LJLIB_MODULE_table
+static const lua_CFunction lj_lib_cf_table[] = {
+ lj_cf_table_foreachi,
+ lj_cf_table_foreach,
+ lj_ffh_table_getn,
+ lj_cf_table_maxn,
+ lj_cf_table_insert,
+ lj_cf_table_remove,
+ lj_cf_table_concat,
+ lj_cf_table_sort
+};
+static const uint8_t lj_lib_init_table[] = {
+92,61,8,8,102,111,114,101,97,99,104,105,7,102,111,114,101,97,99,104,68,103,
+101,116,110,4,109,97,120,110,6,105,110,115,101,114,116,6,114,101,109,111,118,
+101,6,99,111,110,99,97,116,4,115,111,114,116,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_io_method
+#undef LJLIB_MODULE_io_method
+static const lua_CFunction lj_lib_cf_io_method[] = {
+ lj_cf_io_method_close,
+ lj_cf_io_method_read,
+ lj_cf_io_method_write,
+ lj_cf_io_method_flush,
+ lj_cf_io_method_seek,
+ lj_cf_io_method_setvbuf,
+ lj_cf_io_method_lines,
+ lj_cf_io_method___gc,
+ lj_cf_io_method___tostring
+};
+static const uint8_t lj_lib_init_io_method[] = {
+100,62,10,5,99,108,111,115,101,4,114,101,97,100,5,119,114,105,116,101,5,102,
+108,117,115,104,4,115,101,101,107,7,115,101,116,118,98,117,102,5,108,105,110,
+101,115,4,95,95,103,99,10,95,95,116,111,115,116,114,105,110,103,252,1,199,95,
+95,105,110,100,101,120,250,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_io
+#undef LJLIB_MODULE_io
+static const lua_CFunction lj_lib_cf_io[] = {
+ lj_cf_io_open,
+ lj_cf_io_popen,
+ lj_cf_io_tmpfile,
+ lj_cf_io_close,
+ lj_cf_io_read,
+ lj_cf_io_write,
+ lj_cf_io_flush,
+ lj_cf_io_input,
+ lj_cf_io_output,
+ lj_cf_io_lines,
+ lj_cf_io_type
+};
+static const uint8_t lj_lib_init_io[] = {
+109,62,12,252,2,192,250,4,111,112,101,110,5,112,111,112,101,110,7,116,109,112,
+102,105,108,101,5,99,108,111,115,101,4,114,101,97,100,5,119,114,105,116,101,
+5,102,108,117,115,104,5,105,110,112,117,116,6,111,117,116,112,117,116,5,108,
+105,110,101,115,4,116,121,112,101,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_os
+#undef LJLIB_MODULE_os
+static const lua_CFunction lj_lib_cf_os[] = {
+ lj_cf_os_execute,
+ lj_cf_os_remove,
+ lj_cf_os_rename,
+ lj_cf_os_tmpname,
+ lj_cf_os_getenv,
+ lj_cf_os_exit,
+ lj_cf_os_clock,
+ lj_cf_os_date,
+ lj_cf_os_time,
+ lj_cf_os_difftime,
+ lj_cf_os_setlocale
+};
+static const uint8_t lj_lib_init_os[] = {
+120,62,11,7,101,120,101,99,117,116,101,6,114,101,109,111,118,101,6,114,101,
+110,97,109,101,7,116,109,112,110,97,109,101,6,103,101,116,101,110,118,4,101,
+120,105,116,5,99,108,111,99,107,4,100,97,116,101,4,116,105,109,101,8,100,105,
+102,102,116,105,109,101,9,115,101,116,108,111,99,97,108,101,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_debug
+#undef LJLIB_MODULE_debug
+static const lua_CFunction lj_lib_cf_debug[] = {
+ lj_cf_debug_getregistry,
+ lj_cf_debug_getmetatable,
+ lj_cf_debug_setmetatable,
+ lj_cf_debug_getfenv,
+ lj_cf_debug_setfenv,
+ lj_cf_debug_getinfo,
+ lj_cf_debug_getlocal,
+ lj_cf_debug_setlocal,
+ lj_cf_debug_getupvalue,
+ lj_cf_debug_setupvalue,
+ lj_cf_debug_upvalueid,
+ lj_cf_debug_upvaluejoin,
+ lj_cf_debug_sethook,
+ lj_cf_debug_gethook,
+ lj_cf_debug_debug,
+ lj_cf_debug_traceback
+};
+static const uint8_t lj_lib_init_debug[] = {
+131,62,16,11,103,101,116,114,101,103,105,115,116,114,121,12,103,101,116,109,
+101,116,97,116,97,98,108,101,12,115,101,116,109,101,116,97,116,97,98,108,101,
+7,103,101,116,102,101,110,118,7,115,101,116,102,101,110,118,7,103,101,116,105,
+110,102,111,8,103,101,116,108,111,99,97,108,8,115,101,116,108,111,99,97,108,
+10,103,101,116,117,112,118,97,108,117,101,10,115,101,116,117,112,118,97,108,
+117,101,9,117,112,118,97,108,117,101,105,100,11,117,112,118,97,108,117,101,
+106,111,105,110,7,115,101,116,104,111,111,107,7,103,101,116,104,111,111,107,
+5,100,101,98,117,103,9,116,114,97,99,101,98,97,99,107,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_jit
+#undef LJLIB_MODULE_jit
+static const lua_CFunction lj_lib_cf_jit[] = {
+ lj_cf_jit_on,
+ lj_cf_jit_off,
+ lj_cf_jit_flush,
+ lj_cf_jit_status,
+ lj_cf_jit_attach
+};
+static const uint8_t lj_lib_init_jit[] = {
+147,62,9,2,111,110,3,111,102,102,5,102,108,117,115,104,6,115,116,97,116,117,
+115,6,97,116,116,97,99,104,252,5,194,111,115,250,252,4,196,97,114,99,104,250,
+252,3,203,118,101,114,115,105,111,110,95,110,117,109,250,252,2,199,118,101,
+114,115,105,111,110,250,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_jit_util
+#undef LJLIB_MODULE_jit_util
+static const lua_CFunction lj_lib_cf_jit_util[] = {
+ lj_cf_jit_util_funcinfo,
+ lj_cf_jit_util_funcbc,
+ lj_cf_jit_util_funck,
+ lj_cf_jit_util_funcuvname,
+ lj_cf_jit_util_traceinfo,
+ lj_cf_jit_util_traceir,
+ lj_cf_jit_util_tracek,
+ lj_cf_jit_util_tracesnap,
+ lj_cf_jit_util_tracemc,
+ lj_cf_jit_util_traceexitstub,
+ lj_cf_jit_util_ircalladdr
+};
+static const uint8_t lj_lib_init_jit_util[] = {
+152,62,11,8,102,117,110,99,105,110,102,111,6,102,117,110,99,98,99,5,102,117,
+110,99,107,10,102,117,110,99,117,118,110,97,109,101,9,116,114,97,99,101,105,
+110,102,111,7,116,114,97,99,101,105,114,6,116,114,97,99,101,107,9,116,114,97,
+99,101,115,110,97,112,7,116,114,97,99,101,109,99,13,116,114,97,99,101,101,120,
+105,116,115,116,117,98,10,105,114,99,97,108,108,97,100,100,114,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_jit_opt
+#undef LJLIB_MODULE_jit_opt
+static const lua_CFunction lj_lib_cf_jit_opt[] = {
+ lj_cf_jit_opt_start
+};
+static const uint8_t lj_lib_init_jit_opt[] = {
+163,62,1,5,115,116,97,114,116,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_ffi_meta
+#undef LJLIB_MODULE_ffi_meta
+static const lua_CFunction lj_lib_cf_ffi_meta[] = {
+ lj_cf_ffi_meta___index,
+ lj_cf_ffi_meta___newindex,
+ lj_cf_ffi_meta___eq,
+ lj_cf_ffi_meta___len,
+ lj_cf_ffi_meta___lt,
+ lj_cf_ffi_meta___le,
+ lj_cf_ffi_meta___concat,
+ lj_cf_ffi_meta___call,
+ lj_cf_ffi_meta___add,
+ lj_cf_ffi_meta___sub,
+ lj_cf_ffi_meta___mul,
+ lj_cf_ffi_meta___div,
+ lj_cf_ffi_meta___mod,
+ lj_cf_ffi_meta___pow,
+ lj_cf_ffi_meta___unm,
+ lj_cf_ffi_meta___tostring,
+ lj_cf_ffi_meta___pairs,
+ lj_cf_ffi_meta___ipairs
+};
+static const uint8_t lj_lib_init_ffi_meta[] = {
+164,62,19,7,95,95,105,110,100,101,120,10,95,95,110,101,119,105,110,100,101,
+120,4,95,95,101,113,5,95,95,108,101,110,4,95,95,108,116,4,95,95,108,101,8,95,
+95,99,111,110,99,97,116,6,95,95,99,97,108,108,5,95,95,97,100,100,5,95,95,115,
+117,98,5,95,95,109,117,108,5,95,95,100,105,118,5,95,95,109,111,100,5,95,95,
+112,111,119,5,95,95,117,110,109,10,95,95,116,111,115,116,114,105,110,103,7,
+95,95,112,97,105,114,115,8,95,95,105,112,97,105,114,115,195,102,102,105,203,
+95,95,109,101,116,97,116,97,98,108,101,250,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_ffi_clib
+#undef LJLIB_MODULE_ffi_clib
+static const lua_CFunction lj_lib_cf_ffi_clib[] = {
+ lj_cf_ffi_clib___index,
+ lj_cf_ffi_clib___newindex,
+ lj_cf_ffi_clib___gc
+};
+static const uint8_t lj_lib_init_ffi_clib[] = {
+182,62,3,7,95,95,105,110,100,101,120,10,95,95,110,101,119,105,110,100,101,120,
+4,95,95,103,99,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_ffi_callback
+#undef LJLIB_MODULE_ffi_callback
+static const lua_CFunction lj_lib_cf_ffi_callback[] = {
+ lj_cf_ffi_callback_free,
+ lj_cf_ffi_callback_set
+};
+static const uint8_t lj_lib_init_ffi_callback[] = {
+185,62,3,4,102,114,101,101,3,115,101,116,252,1,199,95,95,105,110,100,101,120,
+250,255
+};
+#endif
+
+#ifdef LJLIB_MODULE_ffi
+#undef LJLIB_MODULE_ffi
+static const lua_CFunction lj_lib_cf_ffi[] = {
+ lj_cf_ffi_cdef,
+ lj_cf_ffi_new,
+ lj_cf_ffi_cast,
+ lj_cf_ffi_typeof,
+ lj_cf_ffi_istype,
+ lj_cf_ffi_sizeof,
+ lj_cf_ffi_alignof,
+ lj_cf_ffi_offsetof,
+ lj_cf_ffi_errno,
+ lj_cf_ffi_string,
+ lj_cf_ffi_copy,
+ lj_cf_ffi_fill,
+ lj_cf_ffi_abi,
+ lj_cf_ffi_metatype,
+ lj_cf_ffi_gc,
+ lj_cf_ffi_load
+};
+static const uint8_t lj_lib_init_ffi[] = {
+187,62,22,4,99,100,101,102,3,110,101,119,4,99,97,115,116,6,116,121,112,101,
+111,102,6,105,115,116,121,112,101,6,115,105,122,101,111,102,7,97,108,105,103,
+110,111,102,8,111,102,102,115,101,116,111,102,5,101,114,114,110,111,6,115,116,
+114,105,110,103,4,99,111,112,121,4,102,105,108,108,3,97,98,105,252,8,192,250,
+8,109,101,116,97,116,121,112,101,252,7,192,250,2,103,99,252,5,192,250,4,108,
+111,97,100,252,4,193,67,250,252,3,194,111,115,250,252,2,196,97,114,99,104,250,
+255
+};
+#endif
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/_lj_recdef.h_ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_recdef.h_
new file mode 100644
index 00000000000..2c3c9be63ca
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/_lj_recdef.h_
@@ -0,0 +1,263 @@
+/* This is a generated file. DO NOT EDIT! */
+
+static const uint16_t recff_idmap[] = {
+0,
+0x0100,
+0x0200,
+0x0300,
+0,
+0,
+0x0400,
+0x0500,
+0x0600,
+0x0700,
+0,
+0,
+0x0800,
+0x0900,
+0x0a00,
+0,
+0x0b00,
+0x0c00,
+0x0d00,
+0,
+0x0e00,
+0x0f00,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0x1000,
+0x1100+(IRFPM_FLOOR),
+0x1100+(IRFPM_CEIL),
+0x1200+(IRFPM_SQRT),
+0x1200+(IRFPM_LOG10),
+0x1200+(IRFPM_EXP),
+0x1200+(IRFPM_SIN),
+0x1200+(IRFPM_COS),
+0x1200+(IRFPM_TAN),
+0x1300+(FF_math_asin),
+0x1300+(FF_math_acos),
+0x1300+(FF_math_atan),
+0x1400+(IRCALL_sinh),
+0x1400+(IRCALL_cosh),
+0x1400+(IRCALL_tanh),
+0,
+0x1500,
+0x1600,
+0x1600,
+0x1700,
+0x1800,
+0x1900,
+0,
+0x1a00,
+0x1b00+(IR_MIN),
+0x1b00+(IR_MAX),
+0x1c00,
+0,
+0x1d00+(IR_TOBIT),
+0x1d00+(IR_BNOT),
+0x1d00+(IR_BSWAP),
+0x1e00+(IR_BSHL),
+0x1e00+(IR_BSHR),
+0x1e00+(IR_BSAR),
+0x1e00+(IR_BROL),
+0x1e00+(IR_BROR),
+0x1f00+(IR_BAND),
+0x1f00+(IR_BOR),
+0x1f00+(IR_BXOR),
+0,
+0x2000,
+0x2100+(0),
+0,
+0x2100+(1),
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0x2200,
+0,
+0x2300,
+0x2400,
+0,
+0,
+0,
+0,
+0x2500+(0),
+0x2600+(0),
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0x2500+(GCROOT_IO_OUTPUT),
+0x2600+(GCROOT_IO_OUTPUT),
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0,
+0x2700+(0),
+0x2700+(1),
+0x2800+(MM_eq),
+0x2800+(MM_len),
+0x2800+(MM_lt),
+0x2800+(MM_le),
+0x2800+(MM_concat),
+0x2900,
+0x2800+(MM_add),
+0x2800+(MM_sub),
+0x2800+(MM_mul),
+0x2800+(MM_div),
+0x2800+(MM_mod),
+0x2800+(MM_pow),
+0x2800+(MM_unm),
+0,
+0,
+0,
+0x2a00+(1),
+0x2a00+(0),
+0,
+0,
+0,
+0,
+0x2b00,
+0x2b00,
+0x2c00,
+0x2d00,
+0x2e00+(FF_ffi_sizeof),
+0x2e00+(FF_ffi_alignof),
+0x2e00+(FF_ffi_offsetof),
+0x2f00,
+0x3000,
+0x3100,
+0x3200,
+0x3300,
+0,
+0x3400
+};
+
+static const RecordFunc recff_func[] = {
+recff_nyi,
+recff_c,
+recff_assert,
+recff_type,
+recff_ipairs_aux,
+recff_ipairs,
+recff_getmetatable,
+recff_setmetatable,
+recff_rawget,
+recff_rawset,
+recff_rawequal,
+recff_select,
+recff_tonumber,
+recff_tostring,
+recff_pcall,
+recff_xpcall,
+recff_math_abs,
+recff_math_round,
+recff_math_unary,
+recff_math_atrig,
+recff_math_htrig,
+recff_math_modf,
+recff_math_degrad,
+recff_math_log,
+recff_math_atan2,
+recff_math_pow,
+recff_math_ldexp,
+recff_math_minmax,
+recff_math_random,
+recff_bit_unary,
+recff_bit_shift,
+recff_bit_nary,
+recff_string_len,
+recff_string_range,
+recff_table_getn,
+recff_table_insert,
+recff_table_remove,
+recff_io_write,
+recff_io_flush,
+recff_cdata_index,
+recff_cdata_arith,
+recff_cdata_call,
+recff_clib_index,
+recff_ffi_new,
+recff_ffi_typeof,
+recff_ffi_istype,
+recff_ffi_xof,
+recff_ffi_errno,
+recff_ffi_string,
+recff_ffi_copy,
+recff_ffi_fill,
+recff_ffi_abi,
+recff_ffi_gc
+};
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64.lua
new file mode 100644
index 00000000000..a7173326acb
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64.lua
@@ -0,0 +1,1216 @@
+----------------------------------------------------------------------------
+-- LuaJIT ARM64 disassembler module.
+--
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
+-- Released under the MIT license. See Copyright Notice in luajit.h
+--
+-- Contributed by Djordje Kovacevic and Stefan Pejic from RT-RK.com.
+-- Sponsored by Cisco Systems, Inc.
+----------------------------------------------------------------------------
+-- This is a helper module used by the LuaJIT machine code dumper module.
+--
+-- It disassembles most user-mode AArch64 instructions.
+-- NYI: Advanced SIMD and VFP instructions.
+------------------------------------------------------------------------------
+
+local type = type
+local sub, byte, format = string.sub, string.byte, string.format
+local match, gmatch, gsub = string.match, string.gmatch, string.gsub
+local concat = table.concat
+local bit = require("bit")
+local band, bor, bxor, tohex = bit.band, bit.bor, bit.bxor, bit.tohex
+local lshift, rshift, arshift = bit.lshift, bit.rshift, bit.arshift
+local ror = bit.ror
+
+------------------------------------------------------------------------------
+-- Opcode maps
+------------------------------------------------------------------------------
+
+local map_adr = { -- PC-relative addressing.
+ shift = 31, mask = 1,
+ [0] = "adrDBx", "adrpDBx"
+}
+
+local map_addsubi = { -- Add/subtract immediate.
+ shift = 29, mask = 3,
+ [0] = "add|movDNIg", "adds|cmnD0NIg", "subDNIg", "subs|cmpD0NIg",
+}
+
+local map_logi = { -- Logical immediate.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 22, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "andDNig", "orr|movDN0ig", "eorDNig", "ands|tstD0Nig"
+ },
+ false -- unallocated
+ },
+ {
+ shift = 29, mask = 3,
+ [0] = "andDNig", "orr|movDN0ig", "eorDNig", "ands|tstD0Nig"
+ }
+}
+
+local map_movwi = { -- Move wide immediate.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 22, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "movnDWRg", false, "movz|movDYRg", "movkDWRg"
+ }, false -- unallocated
+ },
+ {
+ shift = 29, mask = 3,
+ [0] = "movnDWRg", false, "movz|movDYRg", "movkDWRg"
+ },
+}
+
+local map_bitf = { -- Bitfield.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 22, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "sbfm|sbfiz|sbfx|asr|sxtw|sxth|sxtbDN12w",
+ "bfm|bfi|bfxilDN13w",
+ "ubfm|ubfiz|ubfx|lsr|lsl|uxth|uxtbDN12w"
+ }
+ },
+ {
+ shift = 22, mask = 1,
+ {
+ shift = 29, mask = 3,
+ [0] = "sbfm|sbfiz|sbfx|asr|sxtw|sxth|sxtbDN12x",
+ "bfm|bfi|bfxilDN13x",
+ "ubfm|ubfiz|ubfx|lsr|lsl|uxth|uxtbDN12x"
+ }
+ }
+}
+
+local map_datai = { -- Data processing - immediate.
+ shift = 23, mask = 7,
+ [0] = map_adr, map_adr, map_addsubi, false,
+ map_logi, map_movwi, map_bitf,
+ {
+ shift = 15, mask = 0x1c0c1,
+ [0] = "extr|rorDNM4w", [0x10080] = "extr|rorDNM4x",
+ [0x10081] = "extr|rorDNM4x"
+ }
+}
+
+local map_logsr = { -- Logical, shifted register.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 15, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = {
+ shift = 21, mask = 7,
+ [0] = "andDNMSg", "bicDNMSg", "andDNMSg", "bicDNMSg",
+ "andDNMSg", "bicDNMSg", "andDNMg", "bicDNMg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] ="orr|movDN0MSg", "orn|mvnDN0MSg", "orr|movDN0MSg", "orn|mvnDN0MSg",
+ "orr|movDN0MSg", "orn|mvnDN0MSg", "orr|movDN0Mg", "orn|mvnDN0Mg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "eorDNMSg", "eonDNMSg", "eorDNMSg", "eonDNMSg",
+ "eorDNMSg", "eonDNMSg", "eorDNMg", "eonDNMg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "ands|tstD0NMSg", "bicsDNMSg", "ands|tstD0NMSg", "bicsDNMSg",
+ "ands|tstD0NMSg", "bicsDNMSg", "ands|tstD0NMg", "bicsDNMg"
+ }
+ },
+ false -- unallocated
+ },
+ {
+ shift = 29, mask = 3,
+ [0] = {
+ shift = 21, mask = 7,
+ [0] = "andDNMSg", "bicDNMSg", "andDNMSg", "bicDNMSg",
+ "andDNMSg", "bicDNMSg", "andDNMg", "bicDNMg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "orr|movDN0MSg", "orn|mvnDN0MSg", "orr|movDN0MSg", "orn|mvnDN0MSg",
+ "orr|movDN0MSg", "orn|mvnDN0MSg", "orr|movDN0Mg", "orn|mvnDN0Mg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "eorDNMSg", "eonDNMSg", "eorDNMSg", "eonDNMSg",
+ "eorDNMSg", "eonDNMSg", "eorDNMg", "eonDNMg"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "ands|tstD0NMSg", "bicsDNMSg", "ands|tstD0NMSg", "bicsDNMSg",
+ "ands|tstD0NMSg", "bicsDNMSg", "ands|tstD0NMg", "bicsDNMg"
+ }
+ }
+}
+
+local map_assh = {
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 15, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "addDNMSg", "addDNMSg", "addDNMSg", "addDNMg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "adds|cmnD0NMSg", "adds|cmnD0NMSg",
+ "adds|cmnD0NMSg", "adds|cmnD0NMg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "sub|negDN0MSg", "sub|negDN0MSg", "sub|negDN0MSg", "sub|negDN0Mg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "subs|cmp|negsD0N0MzSg", "subs|cmp|negsD0N0MzSg",
+ "subs|cmp|negsD0N0MzSg", "subs|cmp|negsD0N0Mzg"
+ },
+ },
+ false -- unallocated
+ },
+ {
+ shift = 29, mask = 3,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "addDNMSg", "addDNMSg", "addDNMSg", "addDNMg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "adds|cmnD0NMSg", "adds|cmnD0NMSg", "adds|cmnD0NMSg",
+ "adds|cmnD0NMg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "sub|negDN0MSg", "sub|negDN0MSg", "sub|negDN0MSg", "sub|negDN0Mg"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "subs|cmp|negsD0N0MzSg", "subs|cmp|negsD0N0MzSg",
+ "subs|cmp|negsD0N0MzSg", "subs|cmp|negsD0N0Mzg"
+ }
+ }
+}
+
+local map_addsubsh = { -- Add/subtract, shifted register.
+ shift = 22, mask = 3,
+ [0] = map_assh, map_assh, map_assh
+}
+
+local map_addsubex = { -- Add/subtract, extended register.
+ shift = 22, mask = 3,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "addDNMXg", "adds|cmnD0NMXg", "subDNMXg", "subs|cmpD0NMzXg",
+ }
+}
+
+local map_addsubc = { -- Add/subtract, with carry.
+ shift = 10, mask = 63,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "adcDNMg", "adcsDNMg", "sbc|ngcDN0Mg", "sbcs|ngcsDN0Mg",
+ }
+}
+
+local map_ccomp = {
+ shift = 4, mask = 1,
+ [0] = {
+ shift = 10, mask = 3,
+ [0] = { -- Conditional compare register.
+ shift = 29, mask = 3,
+ "ccmnNMVCg", false, "ccmpNMVCg",
+ },
+ [2] = { -- Conditional compare immediate.
+ shift = 29, mask = 3,
+ "ccmnN5VCg", false, "ccmpN5VCg",
+ }
+ }
+}
+
+local map_csel = { -- Conditional select.
+ shift = 11, mask = 1,
+ [0] = {
+ shift = 10, mask = 1,
+ [0] = {
+ shift = 29, mask = 3,
+ [0] = "cselDNMzCg", false, "csinv|cinv|csetmDNMcg", false,
+ },
+ {
+ shift = 29, mask = 3,
+ [0] = "csinc|cinc|csetDNMcg", false, "csneg|cnegDNMcg", false,
+ }
+ }
+}
+
+local map_data1s = { -- Data processing, 1 source.
+ shift = 29, mask = 1,
+ [0] = {
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 10, mask = 0x7ff,
+ [0] = "rbitDNg", "rev16DNg", "revDNw", false, "clzDNg", "clsDNg"
+ },
+ {
+ shift = 10, mask = 0x7ff,
+ [0] = "rbitDNg", "rev16DNg", "rev32DNx", "revDNx", "clzDNg", "clsDNg"
+ }
+ }
+}
+
+local map_data2s = { -- Data processing, 2 sources.
+ shift = 29, mask = 1,
+ [0] = {
+ shift = 10, mask = 63,
+ false, "udivDNMg", "sdivDNMg", false, false, false, false, "lslDNMg",
+ "lsrDNMg", "asrDNMg", "rorDNMg"
+ }
+}
+
+local map_data3s = { -- Data processing, 3 sources.
+ shift = 29, mask = 7,
+ [0] = {
+ shift = 21, mask = 7,
+ [0] = {
+ shift = 15, mask = 1,
+ [0] = "madd|mulDNMA0g", "msub|mnegDNMA0g"
+ }
+ }, false, false, false,
+ {
+ shift = 15, mask = 1,
+ [0] = {
+ shift = 21, mask = 7,
+ [0] = "madd|mulDNMA0g", "smaddl|smullDxNMwA0x", "smulhDNMx", false,
+ false, "umaddl|umullDxNMwA0x", "umulhDNMx"
+ },
+ {
+ shift = 21, mask = 7,
+ [0] = "msub|mnegDNMA0g", "smsubl|smneglDxNMwA0x", false, false,
+ false, "umsubl|umneglDxNMwA0x"
+ }
+ }
+}
+
+local map_datar = { -- Data processing, register.
+ shift = 28, mask = 1,
+ [0] = {
+ shift = 24, mask = 1,
+ [0] = map_logsr,
+ {
+ shift = 21, mask = 1,
+ [0] = map_addsubsh, map_addsubex
+ }
+ },
+ {
+ shift = 21, mask = 15,
+ [0] = map_addsubc, false, map_ccomp, false, map_csel, false,
+ {
+ shift = 30, mask = 1,
+ [0] = map_data2s, map_data1s
+ },
+ false, map_data3s, map_data3s, map_data3s, map_data3s, map_data3s,
+ map_data3s, map_data3s, map_data3s
+ }
+}
+
+local map_lrl = { -- Load register, literal.
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 30, mask = 3,
+ [0] = "ldrDwB", "ldrDxB", "ldrswDxB"
+ },
+ {
+ shift = 30, mask = 3,
+ [0] = "ldrDsB", "ldrDdB"
+ }
+}
+
+local map_lsriind = { -- Load/store register, immediate pre/post-indexed.
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strbDwzL", "ldrbDwzL", "ldrsbDxzL", "ldrsbDwzL"
+ }
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strhDwzL", "ldrhDwzL", "ldrshDxzL", "ldrshDwzL"
+ }
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strDwzL", "ldrDwzL", "ldrswDxzL"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDszL", "ldrDszL"
+ }
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strDxzL", "ldrDxzL"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDdzL", "ldrDdzL"
+ }
+ }
+}
+
+local map_lsriro = {
+ shift = 21, mask = 1,
+ [0] = { -- Load/store register immediate.
+ shift = 10, mask = 3,
+ [0] = { -- Unscaled immediate.
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "sturbDwK", "ldurbDwK"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "sturhDwK", "ldurhDwK"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "sturDwK", "ldurDwK"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "sturDxK", "ldurDxK"
+ }
+ }
+ }, map_lsriind, false, map_lsriind
+ },
+ { -- Load/store register, register offset.
+ shift = 10, mask = 3,
+ [2] = {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strbDwO", "ldrbDwO", "ldrsbDxO", "ldrsbDwO"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strhDwO", "ldrhDwO", "ldrshDxO", "ldrshDwO"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDwO", "ldrDwO", "ldrswDxO"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDxO", "ldrDxO"
+ }
+ },
+ {
+ shift = 30, mask = 3,
+ [2] = {
+ shift = 22, mask = 3,
+ [0] = "strDsO", "ldrDsO"
+ },
+ [3] = {
+ shift = 22, mask = 3,
+ [0] = "strDdO", "ldrDdO"
+ }
+ }
+ }
+ }
+}
+
+local map_lsp = { -- Load/store register pair, offset.
+ shift = 22, mask = 1,
+ [0] = {
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 26, mask = 1,
+ [0] = "stpDzAzwP", "stpDzAzsP",
+ },
+ {
+ shift = 26, mask = 1,
+ "stpDzAzdP"
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = "stpDzAzxP"
+ }
+ },
+ {
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 26, mask = 1,
+ [0] = "ldpDzAzwP", "ldpDzAzsP",
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = "ldpswDAxP", "ldpDzAzdP"
+ },
+ {
+ shift = 26, mask = 1,
+ [0] = "ldpDzAzxP"
+ }
+ }
+}
+
+local map_ls = { -- Loads and stores.
+ shift = 24, mask = 0x31,
+ [0x10] = map_lrl, [0x30] = map_lsriro,
+ [0x20] = {
+ shift = 23, mask = 3,
+ map_lsp, map_lsp, map_lsp
+ },
+ [0x21] = {
+ shift = 23, mask = 3,
+ map_lsp, map_lsp, map_lsp
+ },
+ [0x31] = {
+ shift = 26, mask = 1,
+ [0] = {
+ shift = 30, mask = 3,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = "strbDwzU", "ldrbDwzU"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strhDwzU", "ldrhDwzU"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDwzU", "ldrDwzU"
+ },
+ {
+ shift = 22, mask = 3,
+ [0] = "strDxzU", "ldrDxzU"
+ }
+ },
+ {
+ shift = 30, mask = 3,
+ [2] = {
+ shift = 22, mask = 3,
+ [0] = "strDszU", "ldrDszU"
+ },
+ [3] = {
+ shift = 22, mask = 3,
+ [0] = "strDdzU", "ldrDdzU"
+ }
+ }
+ },
+}
+
+local map_datafp = { -- Data processing, SIMD and FP.
+ shift = 28, mask = 7,
+ { -- 001
+ shift = 24, mask = 1,
+ [0] = {
+ shift = 21, mask = 1,
+ {
+ shift = 10, mask = 3,
+ [0] = {
+ shift = 12, mask = 1,
+ [0] = {
+ shift = 13, mask = 1,
+ [0] = {
+ shift = 14, mask = 1,
+ [0] = {
+ shift = 15, mask = 1,
+ [0] = { -- FP/int conversion.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 16, mask = 0xff,
+ [0x20] = "fcvtnsDwNs", [0x21] = "fcvtnuDwNs",
+ [0x22] = "scvtfDsNw", [0x23] = "ucvtfDsNw",
+ [0x24] = "fcvtasDwNs", [0x25] = "fcvtauDwNs",
+ [0x26] = "fmovDwNs", [0x27] = "fmovDsNw",
+ [0x28] = "fcvtpsDwNs", [0x29] = "fcvtpuDwNs",
+ [0x30] = "fcvtmsDwNs", [0x31] = "fcvtmuDwNs",
+ [0x38] = "fcvtzsDwNs", [0x39] = "fcvtzuDwNs",
+ [0x60] = "fcvtnsDwNd", [0x61] = "fcvtnuDwNd",
+ [0x62] = "scvtfDdNw", [0x63] = "ucvtfDdNw",
+ [0x64] = "fcvtasDwNd", [0x65] = "fcvtauDwNd",
+ [0x68] = "fcvtpsDwNd", [0x69] = "fcvtpuDwNd",
+ [0x70] = "fcvtmsDwNd", [0x71] = "fcvtmuDwNd",
+ [0x78] = "fcvtzsDwNd", [0x79] = "fcvtzuDwNd"
+ },
+ {
+ shift = 16, mask = 0xff,
+ [0x20] = "fcvtnsDxNs", [0x21] = "fcvtnuDxNs",
+ [0x22] = "scvtfDsNx", [0x23] = "ucvtfDsNx",
+ [0x24] = "fcvtasDxNs", [0x25] = "fcvtauDxNs",
+ [0x28] = "fcvtpsDxNs", [0x29] = "fcvtpuDxNs",
+ [0x30] = "fcvtmsDxNs", [0x31] = "fcvtmuDxNs",
+ [0x38] = "fcvtzsDxNs", [0x39] = "fcvtzuDxNs",
+ [0x60] = "fcvtnsDxNd", [0x61] = "fcvtnuDxNd",
+ [0x62] = "scvtfDdNx", [0x63] = "ucvtfDdNx",
+ [0x64] = "fcvtasDxNd", [0x65] = "fcvtauDxNd",
+ [0x66] = "fmovDxNd", [0x67] = "fmovDdNx",
+ [0x68] = "fcvtpsDxNd", [0x69] = "fcvtpuDxNd",
+ [0x70] = "fcvtmsDxNd", [0x71] = "fcvtmuDxNd",
+ [0x78] = "fcvtzsDxNd", [0x79] = "fcvtzuDxNd"
+ }
+ }
+ },
+ { -- FP data-processing, 1 source.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 22, mask = 3,
+ [0] = {
+ shift = 15, mask = 63,
+ [0] = "fmovDNf", "fabsDNf", "fnegDNf",
+ "fsqrtDNf", false, "fcvtDdNs", false, false,
+ "frintnDNf", "frintpDNf", "frintmDNf", "frintzDNf",
+ "frintaDNf", false, "frintxDNf", "frintiDNf",
+ },
+ {
+ shift = 15, mask = 63,
+ [0] = "fmovDNf", "fabsDNf", "fnegDNf",
+ "fsqrtDNf", "fcvtDsNd", false, false, false,
+ "frintnDNf", "frintpDNf", "frintmDNf", "frintzDNf",
+ "frintaDNf", false, "frintxDNf", "frintiDNf",
+ }
+ }
+ }
+ },
+ { -- FP compare.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 14, mask = 3,
+ [0] = {
+ shift = 23, mask = 1,
+ [0] = {
+ shift = 0, mask = 31,
+ [0] = "fcmpNMf", [8] = "fcmpNZf",
+ [16] = "fcmpeNMf", [24] = "fcmpeNZf",
+ }
+ }
+ }
+ }
+ },
+ { -- FP immediate.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 5, mask = 31,
+ [0] = {
+ shift = 23, mask = 1,
+ [0] = "fmovDFf"
+ }
+ }
+ }
+ },
+ { -- FP conditional compare.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 23, mask = 1,
+ [0] = {
+ shift = 4, mask = 1,
+ [0] = "fccmpNMVCf", "fccmpeNMVCf"
+ }
+ }
+ },
+ { -- FP data-processing, 2 sources.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 23, mask = 1,
+ [0] = {
+ shift = 12, mask = 15,
+ [0] = "fmulDNMf", "fdivDNMf", "faddDNMf", "fsubDNMf",
+ "fmaxDNMf", "fminDNMf", "fmaxnmDNMf", "fminnmDNMf",
+ "fnmulDNMf"
+ }
+ }
+ },
+ { -- FP conditional select.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 23, mask = 1,
+ [0] = "fcselDNMCf"
+ }
+ }
+ }
+ },
+ { -- FP data-processing, 3 sources.
+ shift = 31, mask = 1,
+ [0] = {
+ shift = 15, mask = 1,
+ [0] = {
+ shift = 21, mask = 5,
+ [0] = "fmaddDNMAf", "fnmaddDNMAf"
+ },
+ {
+ shift = 21, mask = 5,
+ [0] = "fmsubDNMAf", "fnmsubDNMAf"
+ }
+ }
+ }
+ }
+}
+
+local map_br = { -- Branches, exception generating and system instructions.
+ shift = 29, mask = 7,
+ [0] = "bB",
+ { -- Compare & branch, immediate.
+ shift = 24, mask = 3,
+ [0] = "cbzDBg", "cbnzDBg", "tbzDTBw", "tbnzDTBw"
+ },
+ { -- Conditional branch, immediate.
+ shift = 24, mask = 3,
+ [0] = {
+ shift = 4, mask = 1,
+ [0] = {
+ shift = 0, mask = 15,
+ [0] = "beqB", "bneB", "bhsB", "bloB", "bmiB", "bplB", "bvsB", "bvcB",
+ "bhiB", "blsB", "bgeB", "bltB", "bgtB", "bleB", "balB"
+ }
+ }
+ }, false, "blB",
+ { -- Compare & branch, immediate.
+ shift = 24, mask = 3,
+ [0] = "cbzDBg", "cbnzDBg", "tbzDTBx", "tbnzDTBx"
+ },
+ {
+ shift = 24, mask = 3,
+ [0] = { -- Exception generation.
+ shift = 0, mask = 0xe0001f,
+ [0x200000] = "brkW"
+ },
+ { -- System instructions.
+ shift = 0, mask = 0x3fffff,
+ [0x03201f] = "nop"
+ },
+ { -- Unconditional branch, register.
+ shift = 0, mask = 0xfffc1f,
+ [0x1f0000] = "brNx", [0x3f0000] = "blrNx",
+ [0x5f0000] = "retNx"
+ },
+ }
+}
+
+local map_init = {
+ shift = 25, mask = 15,
+ [0] = false, false, false, false, map_ls, map_datar, map_ls, map_datafp,
+ map_datai, map_datai, map_br, map_br, map_ls, map_datar, map_ls, map_datafp
+}
+
+------------------------------------------------------------------------------
+
+local map_regs = { x = {}, w = {}, d = {}, s = {} }
+
+for i=0,30 do
+ map_regs.x[i] = "x"..i
+ map_regs.w[i] = "w"..i
+ map_regs.d[i] = "d"..i
+ map_regs.s[i] = "s"..i
+end
+map_regs.x[31] = "sp"
+map_regs.w[31] = "wsp"
+map_regs.d[31] = "d31"
+map_regs.s[31] = "s31"
+
+local map_cond = {
+ [0] = "eq", "ne", "cs", "cc", "mi", "pl", "vs", "vc",
+ "hi", "ls", "ge", "lt", "gt", "le", "al",
+}
+
+local map_shift = { [0] = "lsl", "lsr", "asr", }
+
+local map_extend = {
+ [0] = "uxtb", "uxth", "uxtw", "uxtx", "sxtb", "sxth", "sxtw", "sxtx",
+}
+
+------------------------------------------------------------------------------
+
+-- Output a nicely formatted line with an opcode and operands.
+local function putop(ctx, text, operands)
+ local pos = ctx.pos
+ local extra = ""
+ if ctx.rel then
+ local sym = ctx.symtab[ctx.rel]
+ if sym then
+ extra = "\t->"..sym
+ end
+ end
+ if ctx.hexdump > 0 then
+ ctx.out(format("%08x %s %-5s %s%s\n",
+ ctx.addr+pos, tohex(ctx.op), text, concat(operands, ", "), extra))
+ else
+ ctx.out(format("%08x %-5s %s%s\n",
+ ctx.addr+pos, text, concat(operands, ", "), extra))
+ end
+ ctx.pos = pos + 4
+end
+
+-- Fallback for unknown opcodes.
+local function unknown(ctx)
+ return putop(ctx, ".long", { "0x"..tohex(ctx.op) })
+end
+
+local function match_reg(p, pat, regnum)
+ return map_regs[match(pat, p.."%w-([xwds])")][regnum]
+end
+
+local function fmt_hex32(x)
+ if x < 0 then
+ return tohex(x)
+ else
+ return format("%x", x)
+ end
+end
+
+local imm13_rep = { 0x55555555, 0x11111111, 0x01010101, 0x00010001, 0x00000001 }
+
+local function decode_imm13(op)
+ local imms = band(rshift(op, 10), 63)
+ local immr = band(rshift(op, 16), 63)
+ if band(op, 0x00400000) == 0 then
+ local len = 5
+ if imms >= 56 then
+ if imms >= 60 then len = 1 else len = 2 end
+ elseif imms >= 48 then len = 3 elseif imms >= 32 then len = 4 end
+ local l = lshift(1, len)-1
+ local s = band(imms, l)
+ local r = band(immr, l)
+ local imm = ror(rshift(-1, 31-s), r)
+ if len ~= 5 then imm = band(imm, lshift(1, l)-1) + rshift(imm, 31-l) end
+ imm = imm * imm13_rep[len]
+ local ix = fmt_hex32(imm)
+ if rshift(op, 31) ~= 0 then
+ return ix..tohex(imm)
+ else
+ return ix
+ end
+ else
+ local lo, hi = -1, 0
+ if imms < 32 then lo = rshift(-1, 31-imms) else hi = rshift(-1, 63-imms) end
+ if immr ~= 0 then
+ lo, hi = ror(lo, immr), ror(hi, immr)
+ local x = immr == 32 and 0 or band(bxor(lo, hi), lshift(-1, 32-immr))
+ lo, hi = bxor(lo, x), bxor(hi, x)
+ if immr >= 32 then lo, hi = hi, lo end
+ end
+ if hi ~= 0 then
+ return fmt_hex32(hi)..tohex(lo)
+ else
+ return fmt_hex32(lo)
+ end
+ end
+end
+
+local function parse_immpc(op, name)
+ if name == "b" or name == "bl" then
+ return arshift(lshift(op, 6), 4)
+ elseif name == "adr" or name == "adrp" then
+ local immlo = band(rshift(op, 29), 3)
+ local immhi = lshift(arshift(lshift(op, 8), 13), 2)
+ return bor(immhi, immlo)
+ elseif name == "tbz" or name == "tbnz" then
+ return lshift(arshift(lshift(op, 13), 18), 2)
+ else
+ return lshift(arshift(lshift(op, 8), 13), 2)
+ end
+end
+
+local function parse_fpimm8(op)
+ local sign = band(op, 0x100000) == 0 and 1 or -1
+ local exp = bxor(rshift(arshift(lshift(op, 12), 5), 24), 0x80) - 131
+ local frac = 16+band(rshift(op, 13), 15)
+ return sign * frac * 2^exp
+end
+
+local function prefer_bfx(sf, uns, imms, immr)
+ if imms < immr or imms == 31 or imms == 63 then
+ return false
+ end
+ if immr == 0 then
+ if sf == 0 and (imms == 7 or imms == 15) then
+ return false
+ end
+ if sf ~= 0 and uns == 0 and (imms == 7 or imms == 15 or imms == 31) then
+ return false
+ end
+ end
+ return true
+end
+
+-- Disassemble a single instruction.
+local function disass_ins(ctx)
+ local pos = ctx.pos
+ local b0, b1, b2, b3 = byte(ctx.code, pos+1, pos+4)
+ local op = bor(lshift(b3, 24), lshift(b2, 16), lshift(b1, 8), b0)
+ local operands = {}
+ local suffix = ""
+ local last, name, pat
+ local map_reg
+ ctx.op = op
+ ctx.rel = nil
+ last = nil
+ local opat
+ opat = map_init[band(rshift(op, 25), 15)]
+ while type(opat) ~= "string" do
+ if not opat then return unknown(ctx) end
+ opat = opat[band(rshift(op, opat.shift), opat.mask)] or opat._
+ end
+ name, pat = match(opat, "^([a-z0-9]*)(.*)")
+ local altname, pat2 = match(pat, "|([a-z0-9_.|]*)(.*)")
+ if altname then pat = pat2 end
+ if sub(pat, 1, 1) == "." then
+ local s2, p2 = match(pat, "^([a-z0-9.]*)(.*)")
+ suffix = suffix..s2
+ pat = p2
+ end
+
+ local rt = match(pat, "[gf]")
+ if rt then
+ if rt == "g" then
+ map_reg = band(op, 0x80000000) ~= 0 and map_regs.x or map_regs.w
+ else
+ map_reg = band(op, 0x400000) ~= 0 and map_regs.d or map_regs.s
+ end
+ end
+
+ local second0, immr
+
+ for p in gmatch(pat, ".") do
+ local x = nil
+ if p == "D" then
+ local regnum = band(op, 31)
+ x = rt and map_reg[regnum] or match_reg(p, pat, regnum)
+ elseif p == "N" then
+ local regnum = band(rshift(op, 5), 31)
+ x = rt and map_reg[regnum] or match_reg(p, pat, regnum)
+ elseif p == "M" then
+ local regnum = band(rshift(op, 16), 31)
+ x = rt and map_reg[regnum] or match_reg(p, pat, regnum)
+ elseif p == "A" then
+ local regnum = band(rshift(op, 10), 31)
+ x = rt and map_reg[regnum] or match_reg(p, pat, regnum)
+ elseif p == "B" then
+ local addr = ctx.addr + pos + parse_immpc(op, name)
+ ctx.rel = addr
+ x = "0x"..tohex(addr)
+ elseif p == "T" then
+ x = bor(band(rshift(op, 26), 32), band(rshift(op, 19), 31))
+ elseif p == "V" then
+ x = band(op, 15)
+ elseif p == "C" then
+ x = map_cond[band(rshift(op, 12), 15)]
+ elseif p == "c" then
+ local rn = band(rshift(op, 5), 31)
+ local rm = band(rshift(op, 16), 31)
+ local cond = band(rshift(op, 12), 15)
+ local invc = bxor(cond, 1)
+ x = map_cond[cond]
+ if altname and cond ~= 14 and cond ~= 15 then
+ local a1, a2 = match(altname, "([^|]*)|(.*)")
+ if rn == rm then
+ local n = #operands
+ operands[n] = nil
+ x = map_cond[invc]
+ if rn ~= 31 then
+ if a1 then name = a1 else name = altname end
+ else
+ operands[n-1] = nil
+ name = a2
+ end
+ end
+ end
+ elseif p == "W" then
+ x = band(rshift(op, 5), 0xffff)
+ elseif p == "Y" then
+ x = band(rshift(op, 5), 0xffff)
+ local hw = band(rshift(op, 21), 3)
+ if altname and (hw == 0 or x ~= 0) then
+ name = altname
+ end
+ elseif p == "L" then
+ local rn = map_regs.x[band(rshift(op, 5), 31)]
+ local imm9 = arshift(lshift(op, 11), 23)
+ if band(op, 0x800) ~= 0 then
+ x = "["..rn..", #"..imm9.."]!"
+ else
+ x = "["..rn.."], #"..imm9
+ end
+ elseif p == "U" then
+ local rn = map_regs.x[band(rshift(op, 5), 31)]
+ local sz = band(rshift(op, 30), 3)
+ local imm12 = lshift(arshift(lshift(op, 10), 20), sz)
+ if imm12 ~= 0 then
+ x = "["..rn..", #"..imm12.."]"
+ else
+ x = "["..rn.."]"
+ end
+ elseif p == "K" then
+ local rn = map_regs.x[band(rshift(op, 5), 31)]
+ local imm9 = arshift(lshift(op, 11), 23)
+ if imm9 ~= 0 then
+ x = "["..rn..", #"..imm9.."]"
+ else
+ x = "["..rn.."]"
+ end
+ elseif p == "O" then
+ local rn, rm = map_regs.x[band(rshift(op, 5), 31)]
+ local m = band(rshift(op, 13), 1)
+ if m == 0 then
+ rm = map_regs.w[band(rshift(op, 16), 31)]
+ else
+ rm = map_regs.x[band(rshift(op, 16), 31)]
+ end
+ x = "["..rn..", "..rm
+ local opt = band(rshift(op, 13), 7)
+ local s = band(rshift(op, 12), 1)
+ local sz = band(rshift(op, 30), 3)
+ -- extension to be applied
+ if opt == 3 then
+ if s == 0 then x = x.."]"
+ else x = x..", lsl #"..sz.."]" end
+ elseif opt == 2 or opt == 6 or opt == 7 then
+ if s == 0 then x = x..", "..map_extend[opt].."]"
+ else x = x..", "..map_extend[opt].." #"..sz.."]" end
+ else
+ x = x.."]"
+ end
+ elseif p == "P" then
+ local opcv, sh = rshift(op, 26), 2
+ if opcv >= 0x2a then sh = 4 elseif opcv >= 0x1b then sh = 3 end
+ local imm7 = lshift(arshift(lshift(op, 10), 25), sh)
+ local rn = map_regs.x[band(rshift(op, 5), 31)]
+ local ind = band(rshift(op, 23), 3)
+ if ind == 1 then
+ x = "["..rn.."], #"..imm7
+ elseif ind == 2 then
+ if imm7 == 0 then
+ x = "["..rn.."]"
+ else
+ x = "["..rn..", #"..imm7.."]"
+ end
+ elseif ind == 3 then
+ x = "["..rn..", #"..imm7.."]!"
+ end
+ elseif p == "I" then
+ local shf = band(rshift(op, 22), 3)
+ local imm12 = band(rshift(op, 10), 0x0fff)
+ local rn, rd = band(rshift(op, 5), 31), band(op, 31)
+ if altname == "mov" and shf == 0 and imm12 == 0 and (rn == 31 or rd == 31) then
+ name = altname
+ x = nil
+ elseif shf == 0 then
+ x = imm12
+ elseif shf == 1 then
+ x = imm12..", lsl #12"
+ end
+ elseif p == "i" then
+ x = "#0x"..decode_imm13(op)
+ elseif p == "1" then
+ immr = band(rshift(op, 16), 63)
+ x = immr
+ elseif p == "2" then
+ x = band(rshift(op, 10), 63)
+ if altname then
+ local a1, a2, a3, a4, a5, a6 =
+ match(altname, "([^|]*)|([^|]*)|([^|]*)|([^|]*)|([^|]*)|(.*)")
+ local sf = band(rshift(op, 26), 32)
+ local uns = band(rshift(op, 30), 1)
+ if prefer_bfx(sf, uns, x, immr) then
+ name = a2
+ x = x - immr + 1
+ elseif immr == 0 and x == 7 then
+ local n = #operands
+ operands[n] = nil
+ if sf ~= 0 then
+ operands[n-1] = gsub(operands[n-1], "x", "w")
+ end
+ last = operands[n-1]
+ name = a6
+ x = nil
+ elseif immr == 0 and x == 15 then
+ local n = #operands
+ operands[n] = nil
+ if sf ~= 0 then
+ operands[n-1] = gsub(operands[n-1], "x", "w")
+ end
+ last = operands[n-1]
+ name = a5
+ x = nil
+ elseif x == 31 or x == 63 then
+ if x == 31 and immr == 0 and name == "sbfm" then
+ name = a4
+ local n = #operands
+ operands[n] = nil
+ if sf ~= 0 then
+ operands[n-1] = gsub(operands[n-1], "x", "w")
+ end
+ last = operands[n-1]
+ else
+ name = a3
+ end
+ x = nil
+ elseif band(x, 31) ~= 31 and immr == x+1 and name == "ubfm" then
+ name = a4
+ last = "#"..(sf+32 - immr)
+ operands[#operands] = last
+ x = nil
+ elseif x < immr then
+ name = a1
+ last = "#"..(sf+32 - immr)
+ operands[#operands] = last
+ x = x + 1
+ end
+ end
+ elseif p == "3" then
+ x = band(rshift(op, 10), 63)
+ if altname then
+ local a1, a2 = match(altname, "([^|]*)|(.*)")
+ if x < immr then
+ name = a1
+ local sf = band(rshift(op, 26), 32)
+ last = "#"..(sf+32 - immr)
+ operands[#operands] = last
+ x = x + 1
+ elseif x >= immr then
+ name = a2
+ x = x - immr + 1
+ end
+ end
+ elseif p == "4" then
+ x = band(rshift(op, 10), 63)
+ local rn = band(rshift(op, 5), 31)
+ local rm = band(rshift(op, 16), 31)
+ if altname and rn == rm then
+ local n = #operands
+ operands[n] = nil
+ last = operands[n-1]
+ name = altname
+ end
+ elseif p == "5" then
+ x = band(rshift(op, 16), 31)
+ elseif p == "S" then
+ x = band(rshift(op, 10), 63)
+ if x == 0 then x = nil
+ else x = map_shift[band(rshift(op, 22), 3)].." #"..x end
+ elseif p == "X" then
+ local opt = band(rshift(op, 13), 7)
+ -- Width specifier <R>.
+ if opt ~= 3 and opt ~= 7 then
+ last = map_regs.w[band(rshift(op, 16), 31)]
+ operands[#operands] = last
+ end
+ x = band(rshift(op, 10), 7)
+ -- Extension.
+ if opt == 2 + band(rshift(op, 31), 1) and
+ band(rshift(op, second0 and 5 or 0), 31) == 31 then
+ if x == 0 then x = nil
+ else x = "lsl #"..x end
+ else
+ if x == 0 then x = map_extend[band(rshift(op, 13), 7)]
+ else x = map_extend[band(rshift(op, 13), 7)].." #"..x end
+ end
+ elseif p == "R" then
+ x = band(rshift(op,21), 3)
+ if x == 0 then x = nil
+ else x = "lsl #"..x*16 end
+ elseif p == "z" then
+ local n = #operands
+ if operands[n] == "sp" then operands[n] = "xzr"
+ elseif operands[n] == "wsp" then operands[n] = "wzr"
+ end
+ elseif p == "Z" then
+ x = 0
+ elseif p == "F" then
+ x = parse_fpimm8(op)
+ elseif p == "g" or p == "f" or p == "x" or p == "w" or
+ p == "d" or p == "s" then
+ -- These are handled in D/N/M/A.
+ elseif p == "0" then
+ if last == "sp" or last == "wsp" then
+ local n = #operands
+ operands[n] = nil
+ last = operands[n-1]
+ if altname then
+ local a1, a2 = match(altname, "([^|]*)|(.*)")
+ if not a1 then
+ name = altname
+ elseif second0 then
+ name, altname = a2, a1
+ else
+ name, altname = a1, a2
+ end
+ end
+ end
+ second0 = true
+ else
+ assert(false)
+ end
+ if x then
+ last = x
+ if type(x) == "number" then x = "#"..x end
+ operands[#operands+1] = x
+ end
+ end
+
+ return putop(ctx, name..suffix, operands)
+end
+
+------------------------------------------------------------------------------
+
+-- Disassemble a block of code.
+local function disass_block(ctx, ofs, len)
+ if not ofs then ofs = 0 end
+ local stop = len and ofs+len or #ctx.code
+ ctx.pos = ofs
+ ctx.rel = nil
+ while ctx.pos < stop do disass_ins(ctx) end
+end
+
+-- Extended API: create a disassembler context. Then call ctx:disass(ofs, len).
+local function create(code, addr, out)
+ local ctx = {}
+ ctx.code = code
+ ctx.addr = addr or 0
+ ctx.out = out or io.write
+ ctx.symtab = {}
+ ctx.disass = disass_block
+ ctx.hexdump = 8
+ return ctx
+end
+
+-- Simple API: disassemble code (a string) at address and output via out.
+local function disass(code, addr, out)
+ create(code, addr, out):disass()
+end
+
+-- Return register name for RID.
+local function regname(r)
+ if r < 32 then return map_regs.x[r] end
+ return map_regs.d[r-32]
+end
+
+-- Public module functions.
+return {
+ create = create,
+ disass = disass,
+ regname = regname
+}
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64be.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64be.lua
new file mode 100644
index 00000000000..7eb389e2fa5
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_arm64be.lua
@@ -0,0 +1,12 @@
+----------------------------------------------------------------------------
+-- LuaJIT ARM64BE disassembler wrapper module.
+--
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
+-- Released under the MIT license. See Copyright Notice in luajit.h
+----------------------------------------------------------------------------
+-- ARM64 instructions are always little-endian. So just forward to the
+-- common ARM64 disassembler module. All the interesting stuff is there.
+------------------------------------------------------------------------------
+
+return require((string.match(..., ".*%.") or "").."dis_arm64")
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64.lua
new file mode 100644
index 00000000000..c4374928ab5
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64.lua
@@ -0,0 +1,17 @@
+----------------------------------------------------------------------------
+-- LuaJIT MIPS64 disassembler wrapper module.
+--
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
+-- Released under the MIT license. See Copyright Notice in luajit.h
+----------------------------------------------------------------------------
+-- This module just exports the big-endian functions from the
+-- MIPS disassembler module. All the interesting stuff is there.
+------------------------------------------------------------------------------
+
+local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
+return {
+ create = dis_mips.create,
+ disass = dis_mips.disass,
+ regname = dis_mips.regname
+}
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64el.lua b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64el.lua
new file mode 100644
index 00000000000..2b1470af505
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/jit/dis_mips64el.lua
@@ -0,0 +1,17 @@
+----------------------------------------------------------------------------
+-- LuaJIT MIPS64EL disassembler wrapper module.
+--
+-- Copyright (C) 2005-2017 Mike Pall. All rights reserved.
+-- Released under the MIT license. See Copyright Notice in luajit.h
+----------------------------------------------------------------------------
+-- This module just exports the little-endian functions from the
+-- MIPS disassembler module. All the interesting stuff is there.
+------------------------------------------------------------------------------
+
+local dis_mips = require((string.match(..., ".*%.") or "").."dis_mips")
+return {
+ create = dis_mips.create_el,
+ disass = dis_mips.disass_el,
+ regname = dis_mips.regname
+}
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h b/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
index dcb5d72346f..29bde38862b 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lauxlib.h
@@ -180,9 +180,13 @@ LUALIB_API void (luaL_addstring) (luaL_Buffer *B, const char *s);
LUALIB_API void (luaL_addvalue) (luaL_Buffer *B);
LUALIB_API void (luaL_pushresult) (luaL_Buffer *B);
+/* -- Luajittex needs this one because it's faster than make it Lua -- */
+LUA_API int (RESERVED_lua_dump) (lua_State *L, lua_Writer writer, void *data, int strip);
/* Luajittex needs this one because it overloads loadfile */
LUALIB_API int (RESERVED_load_aux_JIT) (lua_State *L, int status, int envarg);
+
+
/* }====================================================== */
#endif
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm64.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm64.h
new file mode 100644
index 00000000000..8fd92e76fdd
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_asm_arm64.h
@@ -0,0 +1,2022 @@
+/*
+** ARM64 IR assembler (SSA IR -> machine code).
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
+**
+** Contributed by Djordje Kovacevic and Stefan Pejic from RT-RK.com.
+** Sponsored by Cisco Systems, Inc.
+*/
+
+/* -- Register allocator extensions --------------------------------------- */
+
+/* Allocate a register with a hint. */
+static Reg ra_hintalloc(ASMState *as, IRRef ref, Reg hint, RegSet allow)
+{
+ Reg r = IR(ref)->r;
+ if (ra_noreg(r)) {
+ if (!ra_hashint(r) && !iscrossref(as, ref))
+ ra_sethint(IR(ref)->r, hint); /* Propagate register hint. */
+ r = ra_allocref(as, ref, allow);
+ }
+ ra_noweak(as, r);
+ return r;
+}
+
+/* Allocate two source registers for three-operand instructions. */
+static Reg ra_alloc2(ASMState *as, IRIns *ir, RegSet allow)
+{
+ IRIns *irl = IR(ir->op1), *irr = IR(ir->op2);
+ Reg left = irl->r, right = irr->r;
+ if (ra_hasreg(left)) {
+ ra_noweak(as, left);
+ if (ra_noreg(right))
+ right = ra_allocref(as, ir->op2, rset_exclude(allow, left));
+ else
+ ra_noweak(as, right);
+ } else if (ra_hasreg(right)) {
+ ra_noweak(as, right);
+ left = ra_allocref(as, ir->op1, rset_exclude(allow, right));
+ } else if (ra_hashint(right)) {
+ right = ra_allocref(as, ir->op2, allow);
+ left = ra_alloc1(as, ir->op1, rset_exclude(allow, right));
+ } else {
+ left = ra_allocref(as, ir->op1, allow);
+ right = ra_alloc1(as, ir->op2, rset_exclude(allow, left));
+ }
+ return left | (right << 8);
+}
+
+/* -- Guard handling ------------------------------------------------------ */
+
+/* Setup all needed exit stubs. */
+static void asm_exitstub_setup(ASMState *as, ExitNo nexits)
+{
+ ExitNo i;
+ MCode *mxp = as->mctop;
+ if (mxp - (nexits + 3 + MCLIM_REDZONE) < as->mclim)
+ asm_mclimit(as);
+ /* 1: str lr,[sp]; bl ->vm_exit_handler; movz w0,traceno; bl <1; bl <1; ... */
+ for (i = nexits-1; (int32_t)i >= 0; i--)
+ *--mxp = A64I_LE(A64I_BL|((-3-i)&0x03ffffffu));
+ *--mxp = A64I_LE(A64I_MOVZw|A64F_U16(as->T->traceno));
+ mxp--;
+ *mxp = A64I_LE(A64I_BL|(((MCode *)(void *)lj_vm_exit_handler-mxp)&0x03ffffffu));
+ *--mxp = A64I_LE(A64I_STRx|A64F_D(RID_LR)|A64F_N(RID_SP));
+ as->mctop = mxp;
+}
+
+static MCode *asm_exitstub_addr(ASMState *as, ExitNo exitno)
+{
+ /* Keep this in-sync with exitstub_trace_addr(). */
+ return as->mctop + exitno + 3;
+}
+
+/* Emit conditional branch to exit for guard. */
+static void asm_guardcc(ASMState *as, A64CC cc)
+{
+ MCode *target = asm_exitstub_addr(as, as->snapno);
+ MCode *p = as->mcp;
+ if (LJ_UNLIKELY(p == as->invmcp)) {
+ as->loopinv = 1;
+ *p = A64I_B | ((target-p) & 0x03ffffffu);
+ emit_cond_branch(as, cc^1, p-1);
+ return;
+ }
+ emit_cond_branch(as, cc, target);
+}
+
+/* Emit test and branch instruction to exit for guard. */
+static void asm_guardtnb(ASMState *as, A64Ins ai, Reg r, uint32_t bit)
+{
+ MCode *target = asm_exitstub_addr(as, as->snapno);
+ MCode *p = as->mcp;
+ if (LJ_UNLIKELY(p == as->invmcp)) {
+ as->loopinv = 1;
+ *p = A64I_B | ((target-p) & 0x03ffffffu);
+ emit_tnb(as, ai^0x01000000u, r, bit, p-1);
+ return;
+ }
+ emit_tnb(as, ai, r, bit, target);
+}
+
+/* Emit compare and branch instruction to exit for guard. */
+static void asm_guardcnb(ASMState *as, A64Ins ai, Reg r)
+{
+ MCode *target = asm_exitstub_addr(as, as->snapno);
+ MCode *p = as->mcp;
+ if (LJ_UNLIKELY(p == as->invmcp)) {
+ as->loopinv = 1;
+ *p = A64I_B | ((target-p) & 0x03ffffffu);
+ emit_cnb(as, ai^0x01000000u, r, p-1);
+ return;
+ }
+ emit_cnb(as, ai, r, target);
+}
+
+/* -- Operand fusion ------------------------------------------------------ */
+
+/* Limit linear search to this distance. Avoids O(n^2) behavior. */
+#define CONFLICT_SEARCH_LIM 31
+
+static int asm_isk32(ASMState *as, IRRef ref, int32_t *k)
+{
+ if (irref_isk(ref)) {
+ IRIns *ir = IR(ref);
+ if (ir->o == IR_KNULL || !irt_is64(ir->t)) {
+ *k = ir->i;
+ return 1;
+ } else if (checki32((int64_t)ir_k64(ir)->u64)) {
+ *k = (int32_t)ir_k64(ir)->u64;
+ return 1;
+ }
+ }
+ return 0;
+}
+
+/* Check if there's no conflicting instruction between curins and ref. */
+static int noconflict(ASMState *as, IRRef ref, IROp conflict)
+{
+ IRIns *ir = as->ir;
+ IRRef i = as->curins;
+ if (i > ref + CONFLICT_SEARCH_LIM)
+ return 0; /* Give up, ref is too far away. */
+ while (--i > ref)
+ if (ir[i].o == conflict)
+ return 0; /* Conflict found. */
+ return 1; /* Ok, no conflict. */
+}
+
+/* Fuse the array base of colocated arrays. */
+static int32_t asm_fuseabase(ASMState *as, IRRef ref)
+{
+ IRIns *ir = IR(ref);
+ if (ir->o == IR_TNEW && ir->op1 <= LJ_MAX_COLOSIZE &&
+ !neverfuse(as) && noconflict(as, ref, IR_NEWREF))
+ return (int32_t)sizeof(GCtab);
+ return 0;
+}
+
+#define FUSE_REG 0x40000000
+
+/* Fuse array/hash/upvalue reference into register+offset operand. */
+static Reg asm_fuseahuref(ASMState *as, IRRef ref, int32_t *ofsp, RegSet allow,
+ A64Ins ins)
+{
+ IRIns *ir = IR(ref);
+ if (ra_noreg(ir->r)) {
+ if (ir->o == IR_AREF) {
+ if (mayfuse(as, ref)) {
+ if (irref_isk(ir->op2)) {
+ IRRef tab = IR(ir->op1)->op1;
+ int32_t ofs = asm_fuseabase(as, tab);
+ IRRef refa = ofs ? tab : ir->op1;
+ ofs += 8*IR(ir->op2)->i;
+ if (emit_checkofs(ins, ofs)) {
+ *ofsp = ofs;
+ return ra_alloc1(as, refa, allow);
+ }
+ } else {
+ Reg base = ra_alloc1(as, ir->op1, allow);
+ *ofsp = FUSE_REG|ra_alloc1(as, ir->op2, rset_exclude(allow, base));
+ return base;
+ }
+ }
+ } else if (ir->o == IR_HREFK) {
+ if (mayfuse(as, ref)) {
+ int32_t ofs = (int32_t)(IR(ir->op2)->op2 * sizeof(Node));
+ if (emit_checkofs(ins, ofs)) {
+ *ofsp = ofs;
+ return ra_alloc1(as, ir->op1, allow);
+ }
+ }
+ } else if (ir->o == IR_UREFC) {
+ if (irref_isk(ir->op1)) {
+ GCfunc *fn = ir_kfunc(IR(ir->op1));
+ GCupval *uv = &gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv;
+ int64_t ofs = glofs(as, &uv->tv);
+ if (emit_checkofs(ins, ofs)) {
+ *ofsp = (int32_t)ofs;
+ return RID_GL;
+ }
+ }
+ }
+ }
+ *ofsp = 0;
+ return ra_alloc1(as, ref, allow);
+}
+
+/* Fuse m operand into arithmetic/logic instructions. */
+static uint32_t asm_fuseopm(ASMState *as, A64Ins ai, IRRef ref, RegSet allow)
+{
+ IRIns *ir = IR(ref);
+ if (ra_hasreg(ir->r)) {
+ ra_noweak(as, ir->r);
+ return A64F_M(ir->r);
+ } else if (irref_isk(ref)) {
+ uint32_t m;
+ int64_t k = get_k64val(ir);
+ if ((ai & 0x1f000000) == 0x0a000000)
+ m = emit_isk13(k, irt_is64(ir->t));
+ else
+ m = emit_isk12(k);
+ if (m)
+ return m;
+ } else if (mayfuse(as, ref)) {
+ if ((ir->o >= IR_BSHL && ir->o <= IR_BSAR && irref_isk(ir->op2)) ||
+ (ir->o == IR_ADD && ir->op1 == ir->op2)) {
+ A64Shift sh = ir->o == IR_BSHR ? A64SH_LSR :
+ ir->o == IR_BSAR ? A64SH_ASR : A64SH_LSL;
+ int shift = ir->o == IR_ADD ? 1 :
+ (IR(ir->op2)->i & (irt_is64(ir->t) ? 63 : 31));
+ IRIns *irl = IR(ir->op1);
+ if (sh == A64SH_LSL &&
+ irl->o == IR_CONV &&
+ irl->op2 == ((IRT_I64<<IRCONV_DSH)|IRT_INT|IRCONV_SEXT) &&
+ shift <= 4 &&
+ canfuse(as, irl)) {
+ Reg m = ra_alloc1(as, irl->op1, allow);
+ return A64F_M(m) | A64F_EXSH(A64EX_SXTW, shift);
+ } else {
+ Reg m = ra_alloc1(as, ir->op1, allow);
+ return A64F_M(m) | A64F_SH(sh, shift);
+ }
+ } else if (ir->o == IR_CONV &&
+ ir->op2 == ((IRT_I64<<IRCONV_DSH)|IRT_INT|IRCONV_SEXT)) {
+ Reg m = ra_alloc1(as, ir->op1, allow);
+ return A64F_M(m) | A64F_EX(A64EX_SXTW);
+ }
+ }
+ return A64F_M(ra_allocref(as, ref, allow));
+}
+
+/* Fuse XLOAD/XSTORE reference into load/store operand. */
+static void asm_fusexref(ASMState *as, A64Ins ai, Reg rd, IRRef ref,
+ RegSet allow)
+{
+ IRIns *ir = IR(ref);
+ Reg base;
+ int32_t ofs = 0;
+ if (ra_noreg(ir->r) && canfuse(as, ir)) {
+ if (ir->o == IR_ADD) {
+ if (asm_isk32(as, ir->op2, &ofs) && emit_checkofs(ai, ofs)) {
+ ref = ir->op1;
+ } else {
+ Reg rn, rm;
+ IRRef lref = ir->op1, rref = ir->op2;
+ IRIns *irl = IR(lref);
+ if (mayfuse(as, irl->op1)) {
+ unsigned int shift = 4;
+ if (irl->o == IR_BSHL && irref_isk(irl->op2)) {
+ shift = (IR(irl->op2)->i & 63);
+ } else if (irl->o == IR_ADD && irl->op1 == irl->op2) {
+ shift = 1;
+ }
+ if ((ai >> 30) == shift) {
+ lref = irl->op1;
+ irl = IR(lref);
+ ai |= A64I_LS_SH;
+ }
+ }
+ if (irl->o == IR_CONV &&
+ irl->op2 == ((IRT_I64<<IRCONV_DSH)|IRT_INT|IRCONV_SEXT) &&
+ canfuse(as, irl)) {
+ lref = irl->op1;
+ ai |= A64I_LS_SXTWx;
+ } else {
+ ai |= A64I_LS_LSLx;
+ }
+ rm = ra_alloc1(as, lref, allow);
+ rn = ra_alloc1(as, rref, rset_exclude(allow, rm));
+ emit_dnm(as, (ai^A64I_LS_R), (rd & 31), rn, rm);
+ return;
+ }
+ } else if (ir->o == IR_STRREF) {
+ if (asm_isk32(as, ir->op2, &ofs)) {
+ ref = ir->op1;
+ } else if (asm_isk32(as, ir->op1, &ofs)) {
+ ref = ir->op2;
+ } else {
+ Reg rn = ra_alloc1(as, ir->op1, allow);
+ IRIns *irr = IR(ir->op2);
+ uint32_t m;
+ if (irr+1 == ir && !ra_used(irr) &&
+ irr->o == IR_ADD && irref_isk(irr->op2)) {
+ ofs = sizeof(GCstr) + IR(irr->op2)->i;
+ if (emit_checkofs(ai, ofs)) {
+ Reg rm = ra_alloc1(as, irr->op1, rset_exclude(allow, rn));
+ m = A64F_M(rm) | A64F_EX(A64EX_SXTW);
+ goto skipopm;
+ }
+ }
+ m = asm_fuseopm(as, 0, ir->op2, rset_exclude(allow, rn));
+ ofs = sizeof(GCstr);
+ skipopm:
+ emit_lso(as, ai, rd, rd, ofs);
+ emit_dn(as, A64I_ADDx^m, rd, rn);
+ return;
+ }
+ ofs += sizeof(GCstr);
+ if (!emit_checkofs(ai, ofs)) {
+ Reg rn = ra_alloc1(as, ref, allow);
+ Reg rm = ra_allock(as, ofs, rset_exclude(allow, rn));
+ emit_dnm(as, (ai^A64I_LS_R)|A64I_LS_UXTWx, rd, rn, rm);
+ return;
+ }
+ }
+ }
+ base = ra_alloc1(as, ref, allow);
+ emit_lso(as, ai, (rd & 31), base, ofs);
+}
+
+/* Fuse FP multiply-add/sub. */
+static int asm_fusemadd(ASMState *as, IRIns *ir, A64Ins ai, A64Ins air)
+{
+ IRRef lref = ir->op1, rref = ir->op2;
+ IRIns *irm;
+ if (lref != rref &&
+ ((mayfuse(as, lref) && (irm = IR(lref), irm->o == IR_MUL) &&
+ ra_noreg(irm->r)) ||
+ (mayfuse(as, rref) && (irm = IR(rref), irm->o == IR_MUL) &&
+ (rref = lref, ai = air, ra_noreg(irm->r))))) {
+ Reg dest = ra_dest(as, ir, RSET_FPR);
+ Reg add = ra_hintalloc(as, rref, dest, RSET_FPR);
+ Reg left = ra_alloc2(as, irm,
+ rset_exclude(rset_exclude(RSET_FPR, dest), add));
+ Reg right = (left >> 8); left &= 255;
+ emit_dnma(as, ai, (dest & 31), (left & 31), (right & 31), (add & 31));
+ return 1;
+ }
+ return 0;
+}
+
+/* Fuse BAND + BSHL/BSHR into UBFM. */
+static int asm_fuseandshift(ASMState *as, IRIns *ir)
+{
+ IRIns *irl = IR(ir->op1);
+ lua_assert(ir->o == IR_BAND);
+ if (canfuse(as, irl) && irref_isk(ir->op2)) {
+ uint64_t mask = get_k64val(IR(ir->op2));
+ if (irref_isk(irl->op2) && (irl->o == IR_BSHR || irl->o == IR_BSHL)) {
+ int32_t shmask = irt_is64(irl->t) ? 63 : 31;
+ int32_t shift = (IR(irl->op2)->i & shmask);
+ int32_t imms = shift;
+ if (irl->o == IR_BSHL) {
+ mask >>= shift;
+ shift = (shmask-shift+1) & shmask;
+ imms = 0;
+ }
+ if (mask && !((mask+1) & mask)) { /* Contiguous 1-bits at the bottom. */
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, irl->op1, RSET_GPR);
+ A64Ins ai = shmask == 63 ? A64I_UBFMx : A64I_UBFMw;
+ imms += 63 - emit_clz64(mask);
+ if (imms > shmask) imms = shmask;
+ emit_dn(as, ai | A64F_IMMS(imms) | A64F_IMMR(shift), dest, left);
+ return 1;
+ }
+ }
+ }
+ return 0;
+}
+
+/* Fuse BOR(BSHL, BSHR) into EXTR/ROR. */
+static int asm_fuseorshift(ASMState *as, IRIns *ir)
+{
+ IRIns *irl = IR(ir->op1), *irr = IR(ir->op2);
+ lua_assert(ir->o == IR_BOR);
+ if (canfuse(as, irl) && canfuse(as, irr) &&
+ ((irl->o == IR_BSHR && irr->o == IR_BSHL) ||
+ (irl->o == IR_BSHL && irr->o == IR_BSHR))) {
+ if (irref_isk(irl->op2) && irref_isk(irr->op2)) {
+ IRRef lref = irl->op1, rref = irr->op1;
+ uint32_t lshift = IR(irl->op2)->i, rshift = IR(irr->op2)->i;
+ if (irl->o == IR_BSHR) { /* BSHR needs to be the right operand. */
+ uint32_t tmp2;
+ IRRef tmp1 = lref; lref = rref; rref = tmp1;
+ tmp2 = lshift; lshift = rshift; rshift = tmp2;
+ }
+ if (rshift + lshift == (irt_is64(ir->t) ? 64 : 32)) {
+ A64Ins ai = irt_is64(ir->t) ? A64I_EXTRx : A64I_EXTRw;
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ Reg right = ra_alloc1(as, rref, rset_exclude(RSET_GPR, left));
+ emit_dnm(as, ai | A64F_IMMS(rshift), dest, left, right);
+ return 1;
+ }
+ }
+ }
+ return 0;
+}
+
+/* -- Calls --------------------------------------------------------------- */
+
+/* Generate a call to a C function. */
+static void asm_gencall(ASMState *as, const CCallInfo *ci, IRRef *args)
+{
+ uint32_t n, nargs = CCI_XNARGS(ci);
+ int32_t ofs = 0;
+ Reg gpr, fpr = REGARG_FIRSTFPR;
+ if ((void *)ci->func)
+ emit_call(as, (void *)ci->func);
+ for (gpr = REGARG_FIRSTGPR; gpr <= REGARG_LASTGPR; gpr++)
+ as->cost[gpr] = REGCOST(~0u, ASMREF_L);
+ gpr = REGARG_FIRSTGPR;
+ for (n = 0; n < nargs; n++) { /* Setup args. */
+ IRRef ref = args[n];
+ IRIns *ir = IR(ref);
+ if (ref) {
+ if (irt_isfp(ir->t)) {
+ if (fpr <= REGARG_LASTFPR) {
+ lua_assert(rset_test(as->freeset, fpr)); /* Must have been evicted. */
+ ra_leftov(as, fpr, ref);
+ fpr++;
+ } else {
+ Reg r = ra_alloc1(as, ref, RSET_FPR);
+ emit_spstore(as, ir, r, ofs + ((LJ_BE && !irt_isnum(ir->t)) ? 4 : 0));
+ ofs += 8;
+ }
+ } else {
+ if (gpr <= REGARG_LASTGPR) {
+ lua_assert(rset_test(as->freeset, gpr)); /* Must have been evicted. */
+ ra_leftov(as, gpr, ref);
+ gpr++;
+ } else {
+ Reg r = ra_alloc1(as, ref, RSET_GPR);
+ emit_spstore(as, ir, r, ofs + ((LJ_BE && !irt_is64(ir->t)) ? 4 : 0));
+ ofs += 8;
+ }
+ }
+ }
+ }
+}
+
+/* Setup result reg/sp for call. Evict scratch regs. */
+static void asm_setupresult(ASMState *as, IRIns *ir, const CCallInfo *ci)
+{
+ RegSet drop = RSET_SCRATCH;
+ if (ra_hasreg(ir->r))
+ rset_clear(drop, ir->r); /* Dest reg handled below. */
+ ra_evictset(as, drop); /* Evictions must be performed first. */
+ if (ra_used(ir)) {
+ lua_assert(!irt_ispri(ir->t));
+ if (irt_isfp(ir->t)) {
+ if (ci->flags & CCI_CASTU64) {
+ Reg dest = ra_dest(as, ir, RSET_FPR) & 31;
+ emit_dn(as, irt_isnum(ir->t) ? A64I_FMOV_D_R : A64I_FMOV_S_R,
+ dest, RID_RET);
+ } else {
+ ra_destreg(as, ir, RID_FPRET);
+ }
+ } else {
+ ra_destreg(as, ir, RID_RET);
+ }
+ }
+ UNUSED(ci);
+}
+
+static void asm_callx(ASMState *as, IRIns *ir)
+{
+ IRRef args[CCI_NARGS_MAX*2];
+ CCallInfo ci;
+ IRRef func;
+ IRIns *irf;
+ ci.flags = asm_callx_flags(as, ir);
+ asm_collectargs(as, ir, &ci, args);
+ asm_setupresult(as, ir, &ci);
+ func = ir->op2; irf = IR(func);
+ if (irf->o == IR_CARG) { func = irf->op1; irf = IR(func); }
+ if (irref_isk(func)) { /* Call to constant address. */
+ ci.func = (ASMFunction)(ir_k64(irf)->u64);
+ } else { /* Need a non-argument register for indirect calls. */
+ Reg freg = ra_alloc1(as, func, RSET_RANGE(RID_X8, RID_MAX_GPR)-RSET_FIXED);
+ emit_n(as, A64I_BLR, freg);
+ ci.func = (ASMFunction)(void *)0;
+ }
+ asm_gencall(as, &ci, args);
+}
+
+/* -- Returns ------------------------------------------------------------- */
+
+/* Return to lower frame. Guard that it goes to the right spot. */
+static void asm_retf(ASMState *as, IRIns *ir)
+{
+ Reg base = ra_alloc1(as, REF_BASE, RSET_GPR);
+ void *pc = ir_kptr(IR(ir->op2));
+ int32_t delta = 1+LJ_FR2+bc_a(*((const BCIns *)pc - 1));
+ as->topslot -= (BCReg)delta;
+ if ((int32_t)as->topslot < 0) as->topslot = 0;
+ irt_setmark(IR(REF_BASE)->t); /* Children must not coalesce with BASE reg. */
+ /* Need to force a spill on REF_BASE now to update the stack slot. */
+ emit_lso(as, A64I_STRx, base, RID_SP, ra_spill(as, IR(REF_BASE)));
+ emit_setgl(as, base, jit_base);
+ emit_addptr(as, base, -8*delta);
+ asm_guardcc(as, CC_NE);
+ emit_nm(as, A64I_CMPx, RID_TMP,
+ ra_allock(as, i64ptr(pc), rset_exclude(RSET_GPR, base)));
+ emit_lso(as, A64I_LDRx, RID_TMP, base, -8);
+}
+
+/* -- Type conversions ---------------------------------------------------- */
+
+static void asm_tointg(ASMState *as, IRIns *ir, Reg left)
+{
+ Reg tmp = ra_scratch(as, rset_exclude(RSET_FPR, left));
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ asm_guardcc(as, CC_NE);
+ emit_nm(as, A64I_FCMPd, (tmp & 31), (left & 31));
+ emit_dn(as, A64I_FCVT_F64_S32, (tmp & 31), dest);
+ emit_dn(as, A64I_FCVT_S32_F64, dest, (left & 31));
+}
+
+static void asm_tobit(ASMState *as, IRIns *ir)
+{
+ RegSet allow = RSET_FPR;
+ Reg left = ra_alloc1(as, ir->op1, allow);
+ Reg right = ra_alloc1(as, ir->op2, rset_clear(allow, left));
+ Reg tmp = ra_scratch(as, rset_clear(allow, right));
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ emit_dn(as, A64I_FMOV_R_S, dest, (tmp & 31));
+ emit_dnm(as, A64I_FADDd, (tmp & 31), (left & 31), (right & 31));
+}
+
+static void asm_conv(ASMState *as, IRIns *ir)
+{
+ IRType st = (IRType)(ir->op2 & IRCONV_SRCMASK);
+ int st64 = (st == IRT_I64 || st == IRT_U64 || st == IRT_P64);
+ int stfp = (st == IRT_NUM || st == IRT_FLOAT);
+ IRRef lref = ir->op1;
+ lua_assert(irt_type(ir->t) != st);
+ if (irt_isfp(ir->t)) {
+ Reg dest = ra_dest(as, ir, RSET_FPR);
+ if (stfp) { /* FP to FP conversion. */
+ emit_dn(as, st == IRT_NUM ? A64I_FCVT_F32_F64 : A64I_FCVT_F64_F32,
+ (dest & 31), (ra_alloc1(as, lref, RSET_FPR) & 31));
+ } else { /* Integer to FP conversion. */
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ A64Ins ai = irt_isfloat(ir->t) ?
+ (((IRT_IS64 >> st) & 1) ?
+ (st == IRT_I64 ? A64I_FCVT_F32_S64 : A64I_FCVT_F32_U64) :
+ (st == IRT_INT ? A64I_FCVT_F32_S32 : A64I_FCVT_F32_U32)) :
+ (((IRT_IS64 >> st) & 1) ?
+ (st == IRT_I64 ? A64I_FCVT_F64_S64 : A64I_FCVT_F64_U64) :
+ (st == IRT_INT ? A64I_FCVT_F64_S32 : A64I_FCVT_F64_U32));
+ emit_dn(as, ai, (dest & 31), left);
+ }
+ } else if (stfp) { /* FP to integer conversion. */
+ if (irt_isguard(ir->t)) {
+ /* Checked conversions are only supported from number to int. */
+ lua_assert(irt_isint(ir->t) && st == IRT_NUM);
+ asm_tointg(as, ir, ra_alloc1(as, lref, RSET_FPR));
+ } else {
+ Reg left = ra_alloc1(as, lref, RSET_FPR);
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ A64Ins ai = irt_is64(ir->t) ?
+ (st == IRT_NUM ?
+ (irt_isi64(ir->t) ? A64I_FCVT_S64_F64 : A64I_FCVT_U64_F64) :
+ (irt_isi64(ir->t) ? A64I_FCVT_S64_F32 : A64I_FCVT_U64_F32)) :
+ (st == IRT_NUM ?
+ (irt_isint(ir->t) ? A64I_FCVT_S32_F64 : A64I_FCVT_U32_F64) :
+ (irt_isint(ir->t) ? A64I_FCVT_S32_F32 : A64I_FCVT_U32_F32));
+ emit_dn(as, ai, dest, (left & 31));
+ }
+ } else if (st >= IRT_I8 && st <= IRT_U16) { /* Extend to 32 bit integer. */
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ A64Ins ai = st == IRT_I8 ? A64I_SXTBw :
+ st == IRT_U8 ? A64I_UXTBw :
+ st == IRT_I16 ? A64I_SXTHw : A64I_UXTHw;
+ lua_assert(irt_isint(ir->t) || irt_isu32(ir->t));
+ emit_dn(as, ai, dest, left);
+ } else {
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ if (irt_is64(ir->t)) {
+ if (st64 || !(ir->op2 & IRCONV_SEXT)) {
+ /* 64/64 bit no-op (cast) or 32 to 64 bit zero extension. */
+ ra_leftov(as, dest, lref); /* Do nothing, but may need to move regs. */
+ } else { /* 32 to 64 bit sign extension. */
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ emit_dn(as, A64I_SXTW, dest, left);
+ }
+ } else {
+ if (st64) {
+ /* This is either a 32 bit reg/reg mov which zeroes the hiword
+ ** or a load of the loword from a 64 bit address.
+ */
+ Reg left = ra_alloc1(as, lref, RSET_GPR);
+ emit_dm(as, A64I_MOVw, dest, left);
+ } else { /* 32/32 bit no-op (cast). */
+ ra_leftov(as, dest, lref); /* Do nothing, but may need to move regs. */
+ }
+ }
+ }
+}
+
+static void asm_strto(ASMState *as, IRIns *ir)
+{
+ const CCallInfo *ci = &lj_ir_callinfo[IRCALL_lj_strscan_num];
+ IRRef args[2];
+ Reg dest = 0, tmp;
+ int destused = ra_used(ir);
+ int32_t ofs = 0;
+ ra_evictset(as, RSET_SCRATCH);
+ if (destused) {
+ if (ra_hasspill(ir->s)) {
+ ofs = sps_scale(ir->s);
+ destused = 0;
+ if (ra_hasreg(ir->r)) {
+ ra_free(as, ir->r);
+ ra_modified(as, ir->r);
+ emit_spload(as, ir, ir->r, ofs);
+ }
+ } else {
+ dest = ra_dest(as, ir, RSET_FPR);
+ }
+ }
+ if (destused)
+ emit_lso(as, A64I_LDRd, (dest & 31), RID_SP, 0);
+ asm_guardcnb(as, A64I_CBZ, RID_RET);
+ args[0] = ir->op1; /* GCstr *str */
+ args[1] = ASMREF_TMP1; /* TValue *n */
+ asm_gencall(as, ci, args);
+ tmp = ra_releasetmp(as, ASMREF_TMP1);
+ emit_opk(as, A64I_ADDx, tmp, RID_SP, ofs, RSET_GPR);
+}
+
+/* -- Memory references --------------------------------------------------- */
+
+/* Store tagged value for ref at base+ofs. */
+static void asm_tvstore64(ASMState *as, Reg base, int32_t ofs, IRRef ref)
+{
+ RegSet allow = rset_exclude(RSET_GPR, base);
+ IRIns *ir = IR(ref);
+ lua_assert(irt_ispri(ir->t) || irt_isaddr(ir->t) || irt_isinteger(ir->t));
+ if (irref_isk(ref)) {
+ TValue k;
+ lj_ir_kvalue(as->J->L, &k, ir);
+ emit_lso(as, A64I_STRx, ra_allock(as, k.u64, allow), base, ofs);
+ } else {
+ Reg src = ra_alloc1(as, ref, allow);
+ rset_clear(allow, src);
+ if (irt_isinteger(ir->t)) {
+ Reg type = ra_allock(as, (int64_t)irt_toitype(ir->t) << 47, allow);
+ emit_lso(as, A64I_STRx, RID_TMP, base, ofs);
+ emit_dnm(as, A64I_ADDx | A64F_EX(A64EX_UXTW), RID_TMP, type, src);
+ } else {
+ Reg type = ra_allock(as, (int32_t)irt_toitype(ir->t), allow);
+ emit_lso(as, A64I_STRx, RID_TMP, base, ofs);
+ emit_dnm(as, A64I_ADDx | A64F_SH(A64SH_LSL, 47), RID_TMP, src, type);
+ }
+ }
+}
+
+/* Get pointer to TValue. */
+static void asm_tvptr(ASMState *as, Reg dest, IRRef ref)
+{
+ IRIns *ir = IR(ref);
+ if (irt_isnum(ir->t)) {
+ if (irref_isk(ref)) {
+ /* Use the number constant itself as a TValue. */
+ ra_allockreg(as, i64ptr(ir_knum(ir)), dest);
+ } else {
+ /* Otherwise force a spill and use the spill slot. */
+ emit_opk(as, A64I_ADDx, dest, RID_SP, ra_spill(as, ir), RSET_GPR);
+ }
+ } else {
+ /* Otherwise use g->tmptv to hold the TValue. */
+ asm_tvstore64(as, dest, 0, ref);
+ ra_allockreg(as, i64ptr(&J2G(as->J)->tmptv), dest);
+ }
+}
+
+static void asm_aref(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg idx, base;
+ if (irref_isk(ir->op2)) {
+ IRRef tab = IR(ir->op1)->op1;
+ int32_t ofs = asm_fuseabase(as, tab);
+ IRRef refa = ofs ? tab : ir->op1;
+ uint32_t k = emit_isk12(ofs + 8*IR(ir->op2)->i);
+ if (k) {
+ base = ra_alloc1(as, refa, RSET_GPR);
+ emit_dn(as, A64I_ADDx^k, dest, base);
+ return;
+ }
+ }
+ base = ra_alloc1(as, ir->op1, RSET_GPR);
+ idx = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, base));
+ emit_dnm(as, A64I_ADDx | A64F_EXSH(A64EX_UXTW, 3), dest, base, idx);
+}
+
+/* Inlined hash lookup. Specialized for key type and for const keys.
+** The equivalent C code is:
+** Node *n = hashkey(t, key);
+** do {
+** if (lj_obj_equal(&n->key, key)) return &n->val;
+** } while ((n = nextnode(n)));
+** return niltv(L);
+*/
+static void asm_href(ASMState *as, IRIns *ir, IROp merge)
+{
+ RegSet allow = RSET_GPR;
+ int destused = ra_used(ir);
+ Reg dest = ra_dest(as, ir, allow);
+ Reg tab = ra_alloc1(as, ir->op1, rset_clear(allow, dest));
+ Reg key = 0, tmp = RID_TMP;
+ IRRef refkey = ir->op2;
+ IRIns *irkey = IR(refkey);
+ int isk = irref_isk(ir->op2);
+ IRType1 kt = irkey->t;
+ uint32_t k = 0;
+ uint32_t khash;
+ MCLabel l_end, l_loop, l_next;
+ rset_clear(allow, tab);
+
+ if (!isk) {
+ key = ra_alloc1(as, ir->op2, irt_isnum(kt) ? RSET_FPR : allow);
+ rset_clear(allow, key);
+ if (!irt_isstr(kt)) {
+ tmp = ra_scratch(as, allow);
+ rset_clear(allow, tmp);
+ }
+ } else if (irt_isnum(kt)) {
+ int64_t val = (int64_t)ir_knum(irkey)->u64;
+ if (!(k = emit_isk12(val))) {
+ key = ra_allock(as, val, allow);
+ rset_clear(allow, key);
+ }
+ } else if (!irt_ispri(kt)) {
+ if (!(k = emit_isk12(irkey->i))) {
+ key = ra_alloc1(as, refkey, allow);
+ rset_clear(allow, key);
+ }
+ }
+
+ /* Key not found in chain: jump to exit (if merged) or load niltv. */
+ l_end = emit_label(as);
+ as->invmcp = NULL;
+ if (merge == IR_NE)
+ asm_guardcc(as, CC_AL);
+ else if (destused)
+ emit_loada(as, dest, niltvg(J2G(as->J)));
+
+ /* Follow hash chain until the end. */
+ l_loop = --as->mcp;
+ emit_n(as, A64I_CMPx^A64I_K12^0, dest);
+ emit_lso(as, A64I_LDRx, dest, dest, offsetof(Node, next));
+ l_next = emit_label(as);
+
+ /* Type and value comparison. */
+ if (merge == IR_EQ)
+ asm_guardcc(as, CC_EQ);
+ else
+ emit_cond_branch(as, CC_EQ, l_end);
+
+ if (irt_isnum(kt)) {
+ if (isk) {
+ /* Assumes -0.0 is already canonicalized to +0.0. */
+ if (k)
+ emit_n(as, A64I_CMPx^k, tmp);
+ else
+ emit_nm(as, A64I_CMPx, key, tmp);
+ emit_lso(as, A64I_LDRx, tmp, dest, offsetof(Node, key.u64));
+ } else {
+ Reg tisnum = ra_allock(as, LJ_TISNUM << 15, allow);
+ Reg ftmp = ra_scratch(as, rset_exclude(RSET_FPR, key));
+ rset_clear(allow, tisnum);
+ emit_nm(as, A64I_FCMPd, key, ftmp);
+ emit_dn(as, A64I_FMOV_D_R, (ftmp & 31), (tmp & 31));
+ emit_cond_branch(as, CC_LO, l_next);
+ emit_nm(as, A64I_CMPx | A64F_SH(A64SH_LSR, 32), tisnum, tmp);
+ emit_lso(as, A64I_LDRx, tmp, dest, offsetof(Node, key.n));
+ }
+ } else if (irt_isaddr(kt)) {
+ Reg scr;
+ if (isk) {
+ int64_t kk = ((int64_t)irt_toitype(irkey->t) << 47) | irkey[1].tv.u64;
+ scr = ra_allock(as, kk, allow);
+ emit_nm(as, A64I_CMPx, scr, tmp);
+ emit_lso(as, A64I_LDRx, tmp, dest, offsetof(Node, key.u64));
+ } else {
+ scr = ra_scratch(as, allow);
+ emit_nm(as, A64I_CMPx, tmp, scr);
+ emit_lso(as, A64I_LDRx, scr, dest, offsetof(Node, key.u64));
+ }
+ rset_clear(allow, scr);
+ } else {
+ Reg type, scr;
+ lua_assert(irt_ispri(kt) && !irt_isnil(kt));
+ type = ra_allock(as, ~((int64_t)~irt_toitype(ir->t) << 47), allow);
+ scr = ra_scratch(as, rset_clear(allow, type));
+ rset_clear(allow, scr);
+ emit_nm(as, A64I_CMPw, scr, type);
+ emit_lso(as, A64I_LDRx, scr, dest, offsetof(Node, key));
+ }
+
+ *l_loop = A64I_BCC | A64F_S19(as->mcp - l_loop) | CC_NE;
+ if (!isk && irt_isaddr(kt)) {
+ Reg type = ra_allock(as, (int32_t)irt_toitype(kt), allow);
+ emit_dnm(as, A64I_ADDx | A64F_SH(A64SH_LSL, 47), tmp, key, type);
+ rset_clear(allow, type);
+ }
+ /* Load main position relative to tab->node into dest. */
+ khash = isk ? ir_khash(irkey) : 1;
+ if (khash == 0) {
+ emit_lso(as, A64I_LDRx, dest, tab, offsetof(GCtab, node));
+ } else {
+ emit_dnm(as, A64I_ADDx | A64F_SH(A64SH_LSL, 3), dest, tmp, dest);
+ emit_dnm(as, A64I_ADDx | A64F_SH(A64SH_LSL, 1), dest, dest, dest);
+ emit_lso(as, A64I_LDRx, tmp, tab, offsetof(GCtab, node));
+ if (isk) {
+ Reg tmphash = ra_allock(as, khash, allow);
+ emit_dnm(as, A64I_ANDw, dest, dest, tmphash);
+ emit_lso(as, A64I_LDRw, dest, tab, offsetof(GCtab, hmask));
+ } else if (irt_isstr(kt)) {
+ /* Fetch of str->hash is cheaper than ra_allock. */
+ emit_dnm(as, A64I_ANDw, dest, dest, tmp);
+ emit_lso(as, A64I_LDRw, tmp, key, offsetof(GCstr, hash));
+ emit_lso(as, A64I_LDRw, dest, tab, offsetof(GCtab, hmask));
+ } else { /* Must match with hash*() in lj_tab.c. */
+ emit_dnm(as, A64I_ANDw, dest, dest, tmp);
+ emit_lso(as, A64I_LDRw, tmp, tab, offsetof(GCtab, hmask));
+ emit_dnm(as, A64I_SUBw, dest, dest, tmp);
+ emit_dnm(as, A64I_EXTRw | (A64F_IMMS(32-HASH_ROT3)), tmp, tmp, tmp);
+ emit_dnm(as, A64I_EORw, dest, dest, tmp);
+ emit_dnm(as, A64I_EXTRw | (A64F_IMMS(32-HASH_ROT2)), dest, dest, dest);
+ emit_dnm(as, A64I_SUBw, tmp, tmp, dest);
+ emit_dnm(as, A64I_EXTRw | (A64F_IMMS(32-HASH_ROT1)), dest, dest, dest);
+ emit_dnm(as, A64I_EORw, tmp, tmp, dest);
+ if (irt_isnum(kt)) {
+ emit_dnm(as, A64I_ADDw, dest, dest, dest);
+ emit_dn(as, A64I_LSRx | A64F_IMMR(32)|A64F_IMMS(32), dest, dest);
+ emit_dm(as, A64I_MOVw, tmp, dest);
+ emit_dn(as, A64I_FMOV_R_D, dest, (key & 31));
+ } else {
+ checkmclim(as);
+ emit_dm(as, A64I_MOVw, tmp, key);
+ emit_dnm(as, A64I_EORw, dest, dest,
+ ra_allock(as, irt_toitype(kt) << 15, allow));
+ emit_dn(as, A64I_LSRx | A64F_IMMR(32)|A64F_IMMS(32), dest, dest);
+ emit_dm(as, A64I_MOVx, dest, key);
+ }
+ }
+ }
+}
+
+static void asm_hrefk(ASMState *as, IRIns *ir)
+{
+ IRIns *kslot = IR(ir->op2);
+ IRIns *irkey = IR(kslot->op1);
+ int32_t ofs = (int32_t)(kslot->op2 * sizeof(Node));
+ int32_t kofs = ofs + (int32_t)offsetof(Node, key);
+ int bigofs = !emit_checkofs(A64I_LDRx, ofs);
+ RegSet allow = RSET_GPR;
+ Reg dest = (ra_used(ir) || bigofs) ? ra_dest(as, ir, RSET_GPR) : RID_NONE;
+ Reg node = ra_alloc1(as, ir->op1, allow);
+ Reg key = ra_scratch(as, rset_clear(allow, node));
+ Reg idx = node;
+ uint64_t k;
+ lua_assert(ofs % sizeof(Node) == 0);
+ rset_clear(allow, key);
+ if (bigofs) {
+ idx = dest;
+ rset_clear(allow, dest);
+ kofs = (int32_t)offsetof(Node, key);
+ } else if (ra_hasreg(dest)) {
+ emit_opk(as, A64I_ADDx, dest, node, ofs, allow);
+ }
+ asm_guardcc(as, CC_NE);
+ if (irt_ispri(irkey->t)) {
+ k = ~((int64_t)~irt_toitype(irkey->t) << 47);
+ } else if (irt_isnum(irkey->t)) {
+ k = ir_knum(irkey)->u64;
+ } else {
+ k = ((uint64_t)irt_toitype(irkey->t) << 47) | (uint64_t)ir_kgc(irkey);
+ }
+ emit_nm(as, A64I_CMPx, key, ra_allock(as, k, allow));
+ emit_lso(as, A64I_LDRx, key, idx, kofs);
+ if (bigofs)
+ emit_opk(as, A64I_ADDx, dest, node, ofs, RSET_GPR);
+}
+
+static void asm_uref(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ if (irref_isk(ir->op1)) {
+ GCfunc *fn = ir_kfunc(IR(ir->op1));
+ MRef *v = &gcref(fn->l.uvptr[(ir->op2 >> 8)])->uv.v;
+ emit_lsptr(as, A64I_LDRx, dest, v);
+ } else {
+ Reg uv = ra_scratch(as, RSET_GPR);
+ Reg func = ra_alloc1(as, ir->op1, RSET_GPR);
+ if (ir->o == IR_UREFC) {
+ asm_guardcc(as, CC_NE);
+ emit_n(as, (A64I_CMPx^A64I_K12) | A64F_U12(1), RID_TMP);
+ emit_opk(as, A64I_ADDx, dest, uv,
+ (int32_t)offsetof(GCupval, tv), RSET_GPR);
+ emit_lso(as, A64I_LDRB, RID_TMP, uv, (int32_t)offsetof(GCupval, closed));
+ } else {
+ emit_lso(as, A64I_LDRx, dest, uv, (int32_t)offsetof(GCupval, v));
+ }
+ emit_lso(as, A64I_LDRx, uv, func,
+ (int32_t)offsetof(GCfuncL, uvptr) + 8*(int32_t)(ir->op2 >> 8));
+ }
+}
+
+static void asm_fref(ASMState *as, IRIns *ir)
+{
+ UNUSED(as); UNUSED(ir);
+ lua_assert(!ra_used(ir));
+}
+
+static void asm_strref(ASMState *as, IRIns *ir)
+{
+ RegSet allow = RSET_GPR;
+ Reg dest = ra_dest(as, ir, allow);
+ Reg base = ra_alloc1(as, ir->op1, allow);
+ IRIns *irr = IR(ir->op2);
+ int32_t ofs = sizeof(GCstr);
+ uint32_t m;
+ rset_clear(allow, base);
+ if (irref_isk(ir->op2) && (m = emit_isk12(ofs + irr->i))) {
+ emit_dn(as, A64I_ADDx^m, dest, base);
+ } else {
+ emit_dn(as, (A64I_ADDx^A64I_K12) | A64F_U12(ofs), dest, dest);
+ emit_dnm(as, A64I_ADDx, dest, base, ra_alloc1(as, ir->op2, allow));
+ }
+}
+
+/* -- Loads and stores ---------------------------------------------------- */
+
+static A64Ins asm_fxloadins(IRIns *ir)
+{
+ switch (irt_type(ir->t)) {
+ case IRT_I8: return A64I_LDRB ^ A64I_LS_S;
+ case IRT_U8: return A64I_LDRB;
+ case IRT_I16: return A64I_LDRH ^ A64I_LS_S;
+ case IRT_U16: return A64I_LDRH;
+ case IRT_NUM: return A64I_LDRd;
+ case IRT_FLOAT: return A64I_LDRs;
+ default: return irt_is64(ir->t) ? A64I_LDRx : A64I_LDRw;
+ }
+}
+
+static A64Ins asm_fxstoreins(IRIns *ir)
+{
+ switch (irt_type(ir->t)) {
+ case IRT_I8: case IRT_U8: return A64I_STRB;
+ case IRT_I16: case IRT_U16: return A64I_STRH;
+ case IRT_NUM: return A64I_STRd;
+ case IRT_FLOAT: return A64I_STRs;
+ default: return irt_is64(ir->t) ? A64I_STRx : A64I_STRw;
+ }
+}
+
+static void asm_fload(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg idx;
+ A64Ins ai = asm_fxloadins(ir);
+ int32_t ofs;
+ if (ir->op1 == REF_NIL) {
+ idx = RID_GL;
+ ofs = (ir->op2 << 2) - GG_OFS(g);
+ } else {
+ idx = ra_alloc1(as, ir->op1, RSET_GPR);
+ if (ir->op2 == IRFL_TAB_ARRAY) {
+ ofs = asm_fuseabase(as, ir->op1);
+ if (ofs) { /* Turn the t->array load into an add for colocated arrays. */
+ emit_dn(as, (A64I_ADDx^A64I_K12) | A64F_U12(ofs), dest, idx);
+ return;
+ }
+ }
+ ofs = field_ofs[ir->op2];
+ }
+ emit_lso(as, ai, (dest & 31), idx, ofs);
+}
+
+static void asm_fstore(ASMState *as, IRIns *ir)
+{
+ if (ir->r != RID_SINK) {
+ Reg src = ra_alloc1(as, ir->op2, RSET_GPR);
+ IRIns *irf = IR(ir->op1);
+ Reg idx = ra_alloc1(as, irf->op1, rset_exclude(RSET_GPR, src));
+ int32_t ofs = field_ofs[irf->op2];
+ emit_lso(as, asm_fxstoreins(ir), (src & 31), idx, ofs);
+ }
+}
+
+static void asm_xload(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, irt_isfp(ir->t) ? RSET_FPR : RSET_GPR);
+ lua_assert(!(ir->op2 & IRXLOAD_UNALIGNED));
+ asm_fusexref(as, asm_fxloadins(ir), dest, ir->op1, RSET_GPR);
+}
+
+static void asm_xstore(ASMState *as, IRIns *ir)
+{
+ if (ir->r != RID_SINK) {
+ Reg src = ra_alloc1(as, ir->op2, irt_isfp(ir->t) ? RSET_FPR : RSET_GPR);
+ asm_fusexref(as, asm_fxstoreins(ir), src, ir->op1,
+ rset_exclude(RSET_GPR, src));
+ }
+}
+
+static void asm_ahuvload(ASMState *as, IRIns *ir)
+{
+ Reg idx, tmp, type;
+ int32_t ofs = 0;
+ RegSet gpr = RSET_GPR, allow = irt_isnum(ir->t) ? RSET_FPR : RSET_GPR;
+ lua_assert(irt_isnum(ir->t) || irt_ispri(ir->t) || irt_isaddr(ir->t) ||
+ irt_isint(ir->t));
+ if (ra_used(ir)) {
+ Reg dest = ra_dest(as, ir, allow);
+ tmp = irt_isnum(ir->t) ? ra_scratch(as, rset_clear(gpr, dest)) : dest;
+ if (irt_isaddr(ir->t)) {
+ emit_dn(as, A64I_ANDx^emit_isk13(LJ_GCVMASK, 1), dest, dest);
+ } else if (irt_isnum(ir->t)) {
+ emit_dn(as, A64I_FMOV_D_R, (dest & 31), tmp);
+ } else if (irt_isint(ir->t)) {
+ emit_dm(as, A64I_MOVw, dest, dest);
+ }
+ } else {
+ tmp = ra_scratch(as, gpr);
+ }
+ type = ra_scratch(as, rset_clear(gpr, tmp));
+ idx = asm_fuseahuref(as, ir->op1, &ofs, rset_clear(gpr, type), A64I_LDRx);
+ /* Always do the type check, even if the load result is unused. */
+ asm_guardcc(as, irt_isnum(ir->t) ? CC_LS : CC_NE);
+ if (irt_type(ir->t) >= IRT_NUM) {
+ lua_assert(irt_isinteger(ir->t) || irt_isnum(ir->t));
+ emit_nm(as, A64I_CMPx | A64F_SH(A64SH_LSR, 32),
+ ra_allock(as, LJ_TISNUM << 15, rset_exclude(gpr, idx)), tmp);
+ } else if (irt_isaddr(ir->t)) {
+ emit_n(as, (A64I_CMNx^A64I_K12) | A64F_U12(-irt_toitype(ir->t)), type);
+ emit_dn(as, A64I_ASRx | A64F_IMMR(47), type, tmp);
+ } else if (irt_isnil(ir->t)) {
+ emit_n(as, (A64I_CMNx^A64I_K12) | A64F_U12(1), tmp);
+ } else {
+ emit_nm(as, A64I_CMPx | A64F_SH(A64SH_LSR, 32),
+ ra_allock(as, (irt_toitype(ir->t) << 15) | 0x7fff, allow), tmp);
+ }
+ if (ofs & FUSE_REG)
+ emit_dnm(as, (A64I_LDRx^A64I_LS_R)|A64I_LS_UXTWx|A64I_LS_SH, tmp, idx, (ofs & 31));
+ else
+ emit_lso(as, A64I_LDRx, tmp, idx, ofs);
+}
+
+static void asm_ahustore(ASMState *as, IRIns *ir)
+{
+ if (ir->r != RID_SINK) {
+ RegSet allow = RSET_GPR;
+ Reg idx, src = RID_NONE, tmp = RID_TMP, type = RID_NONE;
+ int32_t ofs = 0;
+ if (irt_isnum(ir->t)) {
+ src = ra_alloc1(as, ir->op2, RSET_FPR);
+ idx = asm_fuseahuref(as, ir->op1, &ofs, allow, A64I_STRd);
+ if (ofs & FUSE_REG)
+ emit_dnm(as, (A64I_STRd^A64I_LS_R)|A64I_LS_UXTWx|A64I_LS_SH, (src & 31), idx, (ofs &31));
+ else
+ emit_lso(as, A64I_STRd, (src & 31), idx, ofs);
+ } else {
+ if (!irt_ispri(ir->t)) {
+ src = ra_alloc1(as, ir->op2, allow);
+ rset_clear(allow, src);
+ if (irt_isinteger(ir->t))
+ type = ra_allock(as, (uint64_t)(int32_t)LJ_TISNUM << 47, allow);
+ else
+ type = ra_allock(as, irt_toitype(ir->t), allow);
+ } else {
+ tmp = type = ra_allock(as, ~((int64_t)~irt_toitype(ir->t)<<47), allow);
+ }
+ idx = asm_fuseahuref(as, ir->op1, &ofs, rset_exclude(allow, type),
+ A64I_STRx);
+ if (ofs & FUSE_REG)
+ emit_dnm(as, (A64I_STRx^A64I_LS_R)|A64I_LS_UXTWx|A64I_LS_SH, tmp, idx, (ofs & 31));
+ else
+ emit_lso(as, A64I_STRx, tmp, idx, ofs);
+ if (ra_hasreg(src)) {
+ if (irt_isinteger(ir->t)) {
+ emit_dnm(as, A64I_ADDx | A64F_EX(A64EX_UXTW), tmp, type, src);
+ } else {
+ emit_dnm(as, A64I_ADDx | A64F_SH(A64SH_LSL, 47), tmp, src, type);
+ }
+ }
+ }
+ }
+}
+
+static void asm_sload(ASMState *as, IRIns *ir)
+{
+ int32_t ofs = 8*((int32_t)ir->op1-2);
+ IRType1 t = ir->t;
+ Reg dest = RID_NONE, base;
+ RegSet allow = RSET_GPR;
+ lua_assert(!(ir->op2 & IRSLOAD_PARENT)); /* Handled by asm_head_side(). */
+ lua_assert(irt_isguard(t) || !(ir->op2 & IRSLOAD_TYPECHECK));
+ if ((ir->op2 & IRSLOAD_CONVERT) && irt_isguard(t) && irt_isint(t)) {
+ dest = ra_scratch(as, RSET_FPR);
+ asm_tointg(as, ir, dest);
+ t.irt = IRT_NUM; /* Continue with a regular number type check. */
+ } else if (ra_used(ir)) {
+ Reg tmp = RID_NONE;
+ if ((ir->op2 & IRSLOAD_CONVERT))
+ tmp = ra_scratch(as, irt_isint(t) ? RSET_FPR : RSET_GPR);
+ lua_assert((irt_isnum(t)) || irt_isint(t) || irt_isaddr(t));
+ dest = ra_dest(as, ir, irt_isnum(t) ? RSET_FPR : allow);
+ base = ra_alloc1(as, REF_BASE, rset_clear(allow, dest));
+ if (irt_isaddr(t)) {
+ emit_dn(as, A64I_ANDx^emit_isk13(LJ_GCVMASK, 1), dest, dest);
+ } else if ((ir->op2 & IRSLOAD_CONVERT)) {
+ if (irt_isint(t)) {
+ emit_dn(as, A64I_FCVT_S32_F64, dest, (tmp & 31));
+ /* If value is already loaded for type check, move it to FPR. */
+ if ((ir->op2 & IRSLOAD_TYPECHECK))
+ emit_dn(as, A64I_FMOV_D_R, (tmp & 31), dest);
+ else
+ dest = tmp;
+ t.irt = IRT_NUM; /* Check for original type. */
+ } else {
+ emit_dn(as, A64I_FCVT_F64_S32, (dest & 31), tmp);
+ dest = tmp;
+ t.irt = IRT_INT; /* Check for original type. */
+ }
+ } else if (irt_isint(t) && (ir->op2 & IRSLOAD_TYPECHECK)) {
+ emit_dm(as, A64I_MOVw, dest, dest);
+ }
+ goto dotypecheck;
+ }
+ base = ra_alloc1(as, REF_BASE, allow);
+dotypecheck:
+ rset_clear(allow, base);
+ if ((ir->op2 & IRSLOAD_TYPECHECK)) {
+ Reg tmp;
+ if (ra_hasreg(dest) && rset_test(RSET_GPR, dest)) {
+ tmp = dest;
+ } else {
+ tmp = ra_scratch(as, allow);
+ rset_clear(allow, tmp);
+ }
+ if (irt_isnum(t) && !(ir->op2 & IRSLOAD_CONVERT))
+ emit_dn(as, A64I_FMOV_D_R, (dest & 31), tmp);
+ /* Need type check, even if the load result is unused. */
+ asm_guardcc(as, irt_isnum(t) ? CC_LS : CC_NE);
+ if (irt_type(t) >= IRT_NUM) {
+ lua_assert(irt_isinteger(t) || irt_isnum(t));
+ emit_nm(as, A64I_CMPx | A64F_SH(A64SH_LSR, 32),
+ ra_allock(as, LJ_TISNUM << 15, allow), tmp);
+ } else if (irt_isnil(t)) {
+ emit_n(as, (A64I_CMNx^A64I_K12) | A64F_U12(1), tmp);
+ } else if (irt_ispri(t)) {
+ emit_nm(as, A64I_CMPx,
+ ra_allock(as, ~((int64_t)~irt_toitype(t) << 47) , allow), tmp);
+ } else {
+ Reg type = ra_scratch(as, allow);
+ emit_n(as, (A64I_CMNx^A64I_K12) | A64F_U12(-irt_toitype(t)), type);
+ emit_dn(as, A64I_ASRx | A64F_IMMR(47), type, tmp);
+ }
+ emit_lso(as, A64I_LDRx, tmp, base, ofs);
+ return;
+ }
+ if (ra_hasreg(dest)) {
+ emit_lso(as, irt_isnum(t) ? A64I_LDRd :
+ (irt_isint(t) ? A64I_LDRw : A64I_LDRx), (dest & 31), base,
+ ofs ^ ((LJ_BE && irt_isint(t) ? 4 : 0)));
+ }
+}
+
+/* -- Allocations --------------------------------------------------------- */
+
+#if LJ_HASFFI
+static void asm_cnew(ASMState *as, IRIns *ir)
+{
+ CTState *cts = ctype_ctsG(J2G(as->J));
+ CTypeID id = (CTypeID)IR(ir->op1)->i;
+ CTSize sz;
+ CTInfo info = lj_ctype_info(cts, id, &sz);
+ const CCallInfo *ci = &lj_ir_callinfo[IRCALL_lj_mem_newgco];
+ IRRef args[4];
+ RegSet allow = (RSET_GPR & ~RSET_SCRATCH);
+ lua_assert(sz != CTSIZE_INVALID || (ir->o == IR_CNEW && ir->op2 != REF_NIL));
+
+ as->gcsteps++;
+ asm_setupresult(as, ir, ci); /* GCcdata * */
+ /* Initialize immutable cdata object. */
+ if (ir->o == IR_CNEWI) {
+ int32_t ofs = sizeof(GCcdata);
+ Reg r = ra_alloc1(as, ir->op2, allow);
+ lua_assert(sz == 4 || sz == 8);
+ emit_lso(as, sz == 8 ? A64I_STRx : A64I_STRw, r, RID_RET, ofs);
+ } else if (ir->op2 != REF_NIL) { /* Create VLA/VLS/aligned cdata. */
+ ci = &lj_ir_callinfo[IRCALL_lj_cdata_newv];
+ args[0] = ASMREF_L; /* lua_State *L */
+ args[1] = ir->op1; /* CTypeID id */
+ args[2] = ir->op2; /* CTSize sz */
+ args[3] = ASMREF_TMP1; /* CTSize align */
+ asm_gencall(as, ci, args);
+ emit_loadi(as, ra_releasetmp(as, ASMREF_TMP1), (int32_t)ctype_align(info));
+ return;
+ }
+
+ /* Initialize gct and ctypeid. lj_mem_newgco() already sets marked. */
+ {
+ Reg r = (id < 65536) ? RID_X1 : ra_allock(as, id, allow);
+ emit_lso(as, A64I_STRB, RID_TMP, RID_RET, offsetof(GCcdata, gct));
+ emit_lso(as, A64I_STRH, r, RID_RET, offsetof(GCcdata, ctypeid));
+ emit_d(as, A64I_MOVZw | A64F_U16(~LJ_TCDATA), RID_TMP);
+ if (id < 65536) emit_d(as, A64I_MOVZw | A64F_U16(id), RID_X1);
+ }
+ args[0] = ASMREF_L; /* lua_State *L */
+ args[1] = ASMREF_TMP1; /* MSize size */
+ asm_gencall(as, ci, args);
+ ra_allockreg(as, (int32_t)(sz+sizeof(GCcdata)),
+ ra_releasetmp(as, ASMREF_TMP1));
+}
+#else
+#define asm_cnew(as, ir) ((void)0)
+#endif
+
+/* -- Write barriers ------------------------------------------------------ */
+
+static void asm_tbar(ASMState *as, IRIns *ir)
+{
+ Reg tab = ra_alloc1(as, ir->op1, RSET_GPR);
+ Reg link = ra_scratch(as, rset_exclude(RSET_GPR, tab));
+ Reg gr = ra_allock(as, i64ptr(J2G(as->J)),
+ rset_exclude(rset_exclude(RSET_GPR, tab), link));
+ Reg mark = RID_TMP;
+ MCLabel l_end = emit_label(as);
+ emit_lso(as, A64I_STRx, link, tab, (int32_t)offsetof(GCtab, gclist));
+ emit_lso(as, A64I_STRB, mark, tab, (int32_t)offsetof(GCtab, marked));
+ emit_lso(as, A64I_STRx, tab, gr,
+ (int32_t)offsetof(global_State, gc.grayagain));
+ emit_dn(as, A64I_ANDw^emit_isk13(~LJ_GC_BLACK, 0), mark, mark);
+ emit_lso(as, A64I_LDRx, link, gr,
+ (int32_t)offsetof(global_State, gc.grayagain));
+ emit_cond_branch(as, CC_EQ, l_end);
+ emit_n(as, A64I_TSTw^emit_isk13(LJ_GC_BLACK, 0), mark);
+ emit_lso(as, A64I_LDRB, mark, tab, (int32_t)offsetof(GCtab, marked));
+}
+
+static void asm_obar(ASMState *as, IRIns *ir)
+{
+ const CCallInfo *ci = &lj_ir_callinfo[IRCALL_lj_gc_barrieruv];
+ IRRef args[2];
+ MCLabel l_end;
+ RegSet allow = RSET_GPR;
+ Reg obj, val, tmp;
+ /* No need for other object barriers (yet). */
+ lua_assert(IR(ir->op1)->o == IR_UREFC);
+ ra_evictset(as, RSET_SCRATCH);
+ l_end = emit_label(as);
+ args[0] = ASMREF_TMP1; /* global_State *g */
+ args[1] = ir->op1; /* TValue *tv */
+ asm_gencall(as, ci, args);
+ ra_allockreg(as, i64ptr(J2G(as->J)), ra_releasetmp(as, ASMREF_TMP1) );
+ obj = IR(ir->op1)->r;
+ tmp = ra_scratch(as, rset_exclude(allow, obj));
+ emit_cond_branch(as, CC_EQ, l_end);
+ emit_n(as, A64I_TSTw^emit_isk13(LJ_GC_BLACK, 0), tmp);
+ emit_cond_branch(as, CC_EQ, l_end);
+ emit_n(as, A64I_TSTw^emit_isk13(LJ_GC_WHITES, 0), RID_TMP);
+ val = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, obj));
+ emit_lso(as, A64I_LDRB, tmp, obj,
+ (int32_t)offsetof(GCupval, marked)-(int32_t)offsetof(GCupval, tv));
+ emit_lso(as, A64I_LDRB, RID_TMP, val, (int32_t)offsetof(GChead, marked));
+}
+
+/* -- Arithmetic and logic operations ------------------------------------- */
+
+static void asm_fparith(ASMState *as, IRIns *ir, A64Ins ai)
+{
+ Reg dest = ra_dest(as, ir, RSET_FPR);
+ Reg right, left = ra_alloc2(as, ir, RSET_FPR);
+ right = (left >> 8); left &= 255;
+ emit_dnm(as, ai, (dest & 31), (left & 31), (right & 31));
+}
+
+static void asm_fpunary(ASMState *as, IRIns *ir, A64Ins ai)
+{
+ Reg dest = ra_dest(as, ir, RSET_FPR);
+ Reg left = ra_hintalloc(as, ir->op1, dest, RSET_FPR);
+ emit_dn(as, ai, (dest & 31), (left & 31));
+}
+
+static void asm_fpmath(ASMState *as, IRIns *ir)
+{
+ IRFPMathOp fpm = (IRFPMathOp)ir->op2;
+ if (fpm == IRFPM_SQRT) {
+ asm_fpunary(as, ir, A64I_FSQRTd);
+ } else if (fpm <= IRFPM_TRUNC) {
+ asm_fpunary(as, ir, fpm == IRFPM_FLOOR ? A64I_FRINTMd :
+ fpm == IRFPM_CEIL ? A64I_FRINTPd : A64I_FRINTZd);
+ } else if (fpm == IRFPM_EXP2 && asm_fpjoin_pow(as, ir)) {
+ return;
+ } else {
+ asm_callid(as, ir, IRCALL_lj_vm_floor + fpm);
+ }
+}
+
+static int asm_swapops(ASMState *as, IRRef lref, IRRef rref)
+{
+ IRIns *ir;
+ if (irref_isk(rref))
+ return 0; /* Don't swap constants to the left. */
+ if (irref_isk(lref))
+ return 1; /* But swap constants to the right. */
+ ir = IR(rref);
+ if ((ir->o >= IR_BSHL && ir->o <= IR_BSAR) ||
+ (ir->o == IR_ADD && ir->op1 == ir->op2) ||
+ (ir->o == IR_CONV && ir->op2 == ((IRT_I64<<IRCONV_DSH)|IRT_INT|IRCONV_SEXT)))
+ return 0; /* Don't swap fusable operands to the left. */
+ ir = IR(lref);
+ if ((ir->o >= IR_BSHL && ir->o <= IR_BSAR) ||
+ (ir->o == IR_ADD && ir->op1 == ir->op2) ||
+ (ir->o == IR_CONV && ir->op2 == ((IRT_I64<<IRCONV_DSH)|IRT_INT|IRCONV_SEXT)))
+ return 1; /* But swap fusable operands to the right. */
+ return 0; /* Otherwise don't swap. */
+}
+
+static void asm_intop(ASMState *as, IRIns *ir, A64Ins ai)
+{
+ IRRef lref = ir->op1, rref = ir->op2;
+ Reg left, dest = ra_dest(as, ir, RSET_GPR);
+ uint32_t m;
+ if ((ai & ~A64I_S) != A64I_SUBw && asm_swapops(as, lref, rref)) {
+ IRRef tmp = lref; lref = rref; rref = tmp;
+ }
+ left = ra_hintalloc(as, lref, dest, RSET_GPR);
+ if (irt_is64(ir->t)) ai |= A64I_X;
+ m = asm_fuseopm(as, ai, rref, rset_exclude(RSET_GPR, left));
+ if (irt_isguard(ir->t)) { /* For IR_ADDOV etc. */
+ asm_guardcc(as, CC_VS);
+ ai |= A64I_S;
+ }
+ emit_dn(as, ai^m, dest, left);
+}
+
+static void asm_intop_s(ASMState *as, IRIns *ir, A64Ins ai)
+{
+ if (as->flagmcp == as->mcp) { /* Drop cmp r, #0. */
+ as->flagmcp = NULL;
+ as->mcp++;
+ ai |= A64I_S;
+ }
+ asm_intop(as, ir, ai);
+}
+
+static void asm_intneg(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_hintalloc(as, ir->op1, dest, RSET_GPR);
+ emit_dm(as, irt_is64(ir->t) ? A64I_NEGx : A64I_NEGw, dest, left);
+}
+
+/* NYI: use add/shift for MUL(OV) with constants. FOLD only does 2^k. */
+static void asm_intmul(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, ir->op1, rset_exclude(RSET_GPR, dest));
+ Reg right = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, left));
+ if (irt_isguard(ir->t)) { /* IR_MULOV */
+ asm_guardcc(as, CC_NE);
+ emit_dm(as, A64I_MOVw, dest, dest); /* Zero-extend. */
+ emit_nm(as, A64I_CMPw | A64F_SH(A64SH_ASR, 31), RID_TMP, dest);
+ emit_dn(as, A64I_ASRx | A64F_IMMR(32), RID_TMP, dest);
+ emit_dnm(as, A64I_SMULL, dest, right, left);
+ } else {
+ emit_dnm(as, irt_is64(ir->t) ? A64I_MULx : A64I_MULw, dest, left, right);
+ }
+}
+
+static void asm_add(ASMState *as, IRIns *ir)
+{
+ if (irt_isnum(ir->t)) {
+ if (!asm_fusemadd(as, ir, A64I_FMADDd, A64I_FMADDd))
+ asm_fparith(as, ir, A64I_FADDd);
+ return;
+ }
+ asm_intop_s(as, ir, A64I_ADDw);
+}
+
+static void asm_sub(ASMState *as, IRIns *ir)
+{
+ if (irt_isnum(ir->t)) {
+ if (!asm_fusemadd(as, ir, A64I_FNMSUBd, A64I_FMSUBd))
+ asm_fparith(as, ir, A64I_FSUBd);
+ return;
+ }
+ asm_intop_s(as, ir, A64I_SUBw);
+}
+
+static void asm_mul(ASMState *as, IRIns *ir)
+{
+ if (irt_isnum(ir->t)) {
+ asm_fparith(as, ir, A64I_FMULd);
+ return;
+ }
+ asm_intmul(as, ir);
+}
+
+static void asm_div(ASMState *as, IRIns *ir)
+{
+#if LJ_HASFFI
+ if (!irt_isnum(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_divi64 :
+ IRCALL_lj_carith_divu64);
+ else
+#endif
+ asm_fparith(as, ir, A64I_FDIVd);
+}
+
+static void asm_pow(ASMState *as, IRIns *ir)
+{
+#if LJ_HASFFI
+ if (!irt_isnum(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_powi64 :
+ IRCALL_lj_carith_powu64);
+ else
+#endif
+ asm_callid(as, ir, IRCALL_lj_vm_powi);
+}
+
+#define asm_addov(as, ir) asm_add(as, ir)
+#define asm_subov(as, ir) asm_sub(as, ir)
+#define asm_mulov(as, ir) asm_mul(as, ir)
+
+#define asm_abs(as, ir) asm_fpunary(as, ir, A64I_FABS)
+#define asm_atan2(as, ir) asm_callid(as, ir, IRCALL_atan2)
+#define asm_ldexp(as, ir) asm_callid(as, ir, IRCALL_ldexp)
+
+static void asm_mod(ASMState *as, IRIns *ir)
+{
+#if LJ_HASFFI
+ if (!irt_isint(ir->t))
+ asm_callid(as, ir, irt_isi64(ir->t) ? IRCALL_lj_carith_modi64 :
+ IRCALL_lj_carith_modu64);
+ else
+#endif
+ asm_callid(as, ir, IRCALL_lj_vm_modi);
+}
+
+static void asm_neg(ASMState *as, IRIns *ir)
+{
+ if (irt_isnum(ir->t)) {
+ asm_fpunary(as, ir, A64I_FNEGd);
+ return;
+ }
+ asm_intneg(as, ir);
+}
+
+static void asm_band(ASMState *as, IRIns *ir)
+{
+ A64Ins ai = A64I_ANDw;
+ if (asm_fuseandshift(as, ir))
+ return;
+ if (as->flagmcp == as->mcp) {
+ /* Try to drop cmp r, #0. */
+ as->flagmcp = NULL;
+ as->mcp++;
+ ai = A64I_ANDSw;
+ }
+ asm_intop(as, ir, ai);
+}
+
+static void asm_borbxor(ASMState *as, IRIns *ir, A64Ins ai)
+{
+ IRRef lref = ir->op1, rref = ir->op2;
+ IRIns *irl = IR(lref), *irr = IR(rref);
+ if ((canfuse(as, irl) && irl->o == IR_BNOT && !irref_isk(rref)) ||
+ (canfuse(as, irr) && irr->o == IR_BNOT && !irref_isk(lref))) {
+ Reg left, dest = ra_dest(as, ir, RSET_GPR);
+ uint32_t m;
+ if (irl->o == IR_BNOT) {
+ IRRef tmp = lref; lref = rref; rref = tmp;
+ }
+ left = ra_alloc1(as, lref, RSET_GPR);
+ ai |= A64I_ON;
+ if (irt_is64(ir->t)) ai |= A64I_X;
+ m = asm_fuseopm(as, ai, IR(rref)->op1, rset_exclude(RSET_GPR, left));
+ emit_dn(as, ai^m, dest, left);
+ } else {
+ asm_intop(as, ir, ai);
+ }
+}
+
+static void asm_bor(ASMState *as, IRIns *ir)
+{
+ if (asm_fuseorshift(as, ir))
+ return;
+ asm_borbxor(as, ir, A64I_ORRw);
+}
+
+#define asm_bxor(as, ir) asm_borbxor(as, ir, A64I_EORw)
+
+static void asm_bnot(ASMState *as, IRIns *ir)
+{
+ A64Ins ai = A64I_MVNw;
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ uint32_t m = asm_fuseopm(as, ai, ir->op1, RSET_GPR);
+ if (irt_is64(ir->t)) ai |= A64I_X;
+ emit_d(as, ai^m, dest);
+}
+
+static void asm_bswap(ASMState *as, IRIns *ir)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, ir->op1, RSET_GPR);
+ emit_dn(as, irt_is64(ir->t) ? A64I_REVx : A64I_REVw, dest, left);
+}
+
+static void asm_bitshift(ASMState *as, IRIns *ir, A64Ins ai, A64Shift sh)
+{
+ int32_t shmask = irt_is64(ir->t) ? 63 : 31;
+ if (irref_isk(ir->op2)) { /* Constant shifts. */
+ Reg left, dest = ra_dest(as, ir, RSET_GPR);
+ int32_t shift = (IR(ir->op2)->i & shmask);
+ IRIns *irl = IR(ir->op1);
+ if (shmask == 63) ai += A64I_UBFMx - A64I_UBFMw;
+
+ /* Fuse BSHL + BSHR/BSAR into UBFM/SBFM aka UBFX/SBFX/UBFIZ/SBFIZ. */
+ if ((sh == A64SH_LSR || sh == A64SH_ASR) && canfuse(as, irl)) {
+ if (irl->o == IR_BSHL && irref_isk(irl->op2)) {
+ int32_t shift2 = (IR(irl->op2)->i & shmask);
+ shift = ((shift - shift2) & shmask);
+ shmask -= shift2;
+ ir = irl;
+ }
+ }
+
+ left = ra_alloc1(as, ir->op1, RSET_GPR);
+ switch (sh) {
+ case A64SH_LSL:
+ emit_dn(as, ai | A64F_IMMS(shmask-shift) |
+ A64F_IMMR((shmask-shift+1)&shmask), dest, left);
+ break;
+ case A64SH_LSR: case A64SH_ASR:
+ emit_dn(as, ai | A64F_IMMS(shmask) | A64F_IMMR(shift), dest, left);
+ break;
+ case A64SH_ROR:
+ emit_dnm(as, ai | A64F_IMMS(shift), dest, left, left);
+ break;
+ }
+ } else { /* Variable-length shifts. */
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_alloc1(as, ir->op1, RSET_GPR);
+ Reg right = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, left));
+ emit_dnm(as, (shmask == 63 ? A64I_SHRx : A64I_SHRw) | A64F_BSH(sh), dest, left, right);
+ }
+}
+
+#define asm_bshl(as, ir) asm_bitshift(as, ir, A64I_UBFMw, A64SH_LSL)
+#define asm_bshr(as, ir) asm_bitshift(as, ir, A64I_UBFMw, A64SH_LSR)
+#define asm_bsar(as, ir) asm_bitshift(as, ir, A64I_SBFMw, A64SH_ASR)
+#define asm_bror(as, ir) asm_bitshift(as, ir, A64I_EXTRw, A64SH_ROR)
+#define asm_brol(as, ir) lua_assert(0)
+
+static void asm_intmin_max(ASMState *as, IRIns *ir, A64CC cc)
+{
+ Reg dest = ra_dest(as, ir, RSET_GPR);
+ Reg left = ra_hintalloc(as, ir->op1, dest, RSET_GPR);
+ Reg right = ra_alloc1(as, ir->op2, rset_exclude(RSET_GPR, left));
+ emit_dnm(as, A64I_CSELw|A64F_CC(cc), dest, left, right);
+ emit_nm(as, A64I_CMPw, left, right);
+}
+
+static void asm_fpmin_max(ASMState *as, IRIns *ir, A64CC fcc)
+{
+ Reg dest = (ra_dest(as, ir, RSET_FPR) & 31);
+ Reg right, left = ra_alloc2(as, ir, RSET_FPR);
+ right = ((left >> 8) & 31); left &= 31;
+ emit_dnm(as, A64I_FCSELd | A64F_CC(fcc), dest, left, right);
+ emit_nm(as, A64I_FCMPd, left, right);
+}
+
+static void asm_min_max(ASMState *as, IRIns *ir, A64CC cc, A64CC fcc)
+{
+ if (irt_isnum(ir->t))
+ asm_fpmin_max(as, ir, fcc);
+ else
+ asm_intmin_max(as, ir, cc);
+}
+
+#define asm_max(as, ir) asm_min_max(as, ir, CC_GT, CC_HI)
+#define asm_min(as, ir) asm_min_max(as, ir, CC_LT, CC_LO)
+
+/* -- Comparisons --------------------------------------------------------- */
+
+/* Map of comparisons to flags. ORDER IR. */
+static const uint8_t asm_compmap[IR_ABC+1] = {
+ /* op FP swp int cc FP cc */
+ /* LT */ CC_GE + (CC_HS << 4),
+ /* GE x */ CC_LT + (CC_HI << 4),
+ /* LE */ CC_GT + (CC_HI << 4),
+ /* GT x */ CC_LE + (CC_HS << 4),
+ /* ULT x */ CC_HS + (CC_LS << 4),
+ /* UGE */ CC_LO + (CC_LO << 4),
+ /* ULE x */ CC_HI + (CC_LO << 4),
+ /* UGT */ CC_LS + (CC_LS << 4),
+ /* EQ */ CC_NE + (CC_NE << 4),
+ /* NE */ CC_EQ + (CC_EQ << 4),
+ /* ABC */ CC_LS + (CC_LS << 4) /* Same as UGT. */
+};
+
+/* FP comparisons. */
+static void asm_fpcomp(ASMState *as, IRIns *ir)
+{
+ Reg left, right;
+ A64Ins ai;
+ int swp = ((ir->o ^ (ir->o >> 2)) & ~(ir->o >> 3) & 1);
+ if (!swp && irref_isk(ir->op2) && ir_knum(IR(ir->op2))->u64 == 0) {
+ left = (ra_alloc1(as, ir->op1, RSET_FPR) & 31);
+ right = 0;
+ ai = A64I_FCMPZd;
+ } else {
+ left = ra_alloc2(as, ir, RSET_FPR);
+ if (swp) {
+ right = (left & 31); left = ((left >> 8) & 31);
+ } else {
+ right = ((left >> 8) & 31); left &= 31;
+ }
+ ai = A64I_FCMPd;
+ }
+ asm_guardcc(as, (asm_compmap[ir->o] >> 4));
+ emit_nm(as, ai, left, right);
+}
+
+/* Integer comparisons. */
+static void asm_intcomp(ASMState *as, IRIns *ir)
+{
+ A64CC oldcc, cc = (asm_compmap[ir->o] & 15);
+ A64Ins ai = irt_is64(ir->t) ? A64I_CMPx : A64I_CMPw;
+ IRRef lref = ir->op1, rref = ir->op2;
+ Reg left;
+ uint32_t m;
+ int cmpprev0 = 0;
+ lua_assert(irt_is64(ir->t) || irt_isint(ir->t) ||
+ irt_isu32(ir->t) || irt_isaddr(ir->t) || irt_isu8(ir->t));
+ if (asm_swapops(as, lref, rref)) {
+ IRRef tmp = lref; lref = rref; rref = tmp;
+ if (cc >= CC_GE) cc ^= 7; /* LT <-> GT, LE <-> GE */
+ else if (cc > CC_NE) cc ^= 11; /* LO <-> HI, LS <-> HS */
+ }
+ oldcc = cc;
+ if (irref_isk(rref) && get_k64val(IR(rref)) == 0) {
+ IRIns *irl = IR(lref);
+ if (cc == CC_GE) cc = CC_PL;
+ else if (cc == CC_LT) cc = CC_MI;
+ else if (cc > CC_NE) goto nocombine; /* Other conds don't work with tst. */
+ cmpprev0 = (irl+1 == ir);
+ /* Combine and-cmp-bcc into tbz/tbnz or and-cmp into tst. */
+ if (cmpprev0 && irl->o == IR_BAND && !ra_used(irl)) {
+ IRRef blref = irl->op1, brref = irl->op2;
+ uint32_t m2 = 0;
+ Reg bleft;
+ if (asm_swapops(as, blref, brref)) {
+ Reg tmp = blref; blref = brref; brref = tmp;
+ }
+ if (irref_isk(brref)) {
+ uint64_t k = get_k64val(IR(brref));
+ if (k && !(k & (k-1)) && (cc == CC_EQ || cc == CC_NE)) {
+ asm_guardtnb(as, cc == CC_EQ ? A64I_TBZ : A64I_TBNZ,
+ ra_alloc1(as, blref, RSET_GPR), emit_ctz64(k));
+ return;
+ }
+ m2 = emit_isk13(k, irt_is64(irl->t));
+ }
+ bleft = ra_alloc1(as, blref, RSET_GPR);
+ ai = (irt_is64(irl->t) ? A64I_TSTx : A64I_TSTw);
+ if (!m2)
+ m2 = asm_fuseopm(as, ai, brref, rset_exclude(RSET_GPR, bleft));
+ asm_guardcc(as, cc);
+ emit_n(as, ai^m2, bleft);
+ return;
+ }
+ if (cc == CC_EQ || cc == CC_NE) {
+ /* Combine cmp-bcc into cbz/cbnz. */
+ ai = cc == CC_EQ ? A64I_CBZ : A64I_CBNZ;
+ if (irt_is64(ir->t)) ai |= A64I_X;
+ asm_guardcnb(as, ai, ra_alloc1(as, lref, RSET_GPR));
+ return;
+ }
+ }
+nocombine:
+ left = ra_alloc1(as, lref, RSET_GPR);
+ m = asm_fuseopm(as, ai, rref, rset_exclude(RSET_GPR, left));
+ asm_guardcc(as, cc);
+ emit_n(as, ai^m, left);
+ /* Signed comparison with zero and referencing previous ins? */
+ if (cmpprev0 && (oldcc <= CC_NE || oldcc >= CC_GE))
+ as->flagmcp = as->mcp; /* Allow elimination of the compare. */
+}
+
+static void asm_comp(ASMState *as, IRIns *ir)
+{
+ if (irt_isnum(ir->t))
+ asm_fpcomp(as, ir);
+ else
+ asm_intcomp(as, ir);
+}
+
+#define asm_equal(as, ir) asm_comp(as, ir)
+
+/* -- Support for 64 bit ops in 32 bit mode ------------------------------- */
+
+/* Hiword op of a split 64 bit op. Previous op must be the loword op. */
+static void asm_hiop(ASMState *as, IRIns *ir)
+{
+ UNUSED(as); UNUSED(ir); lua_assert(0); /* Unused on 64 bit. */
+}
+
+/* -- Profiling ----------------------------------------------------------- */
+
+static void asm_prof(ASMState *as, IRIns *ir)
+{
+ uint32_t k = emit_isk13(HOOK_PROFILE, 0);
+ lua_assert(k != 0);
+ UNUSED(ir);
+ asm_guardcc(as, CC_NE);
+ emit_n(as, A64I_TSTw^k, RID_TMP);
+ emit_lsptr(as, A64I_LDRB, RID_TMP, (void *)&J2G(as->J)->hookmask);
+}
+
+/* -- Stack handling ------------------------------------------------------ */
+
+/* Check Lua stack size for overflow. Use exit handler as fallback. */
+static void asm_stack_check(ASMState *as, BCReg topslot,
+ IRIns *irp, RegSet allow, ExitNo exitno)
+{
+ Reg pbase;
+ uint32_t k;
+ if (irp) {
+ if (!ra_hasspill(irp->s)) {
+ pbase = irp->r;
+ lua_assert(ra_hasreg(pbase));
+ } else if (allow) {
+ pbase = rset_pickbot(allow);
+ } else {
+ pbase = RID_RET;
+ emit_lso(as, A64I_LDRx, RID_RET, RID_SP, 0); /* Restore temp register. */
+ }
+ } else {
+ pbase = RID_BASE;
+ }
+ emit_cond_branch(as, CC_LS, asm_exitstub_addr(as, exitno));
+ k = emit_isk12((8*topslot));
+ lua_assert(k);
+ emit_n(as, A64I_CMPx^k, RID_TMP);
+ emit_dnm(as, A64I_SUBx, RID_TMP, RID_TMP, pbase);
+ emit_lso(as, A64I_LDRx, RID_TMP, RID_TMP,
+ (int32_t)offsetof(lua_State, maxstack));
+ if (irp) { /* Must not spill arbitrary registers in head of side trace. */
+ if (ra_hasspill(irp->s))
+ emit_lso(as, A64I_LDRx, pbase, RID_SP, sps_scale(irp->s));
+ emit_lso(as, A64I_LDRx, RID_TMP, RID_GL, glofs(as, &J2G(as->J)->cur_L));
+ if (ra_hasspill(irp->s) && !allow)
+ emit_lso(as, A64I_STRx, RID_RET, RID_SP, 0); /* Save temp register. */
+ } else {
+ emit_getgl(as, RID_TMP, cur_L);
+ }
+}
+
+/* Restore Lua stack from on-trace state. */
+static void asm_stack_restore(ASMState *as, SnapShot *snap)
+{
+ SnapEntry *map = &as->T->snapmap[snap->mapofs];
+#ifdef LUA_USE_ASSERT
+ SnapEntry *flinks = &as->T->snapmap[snap_nextofs(as->T, snap)-1-LJ_FR2];
+#endif
+ MSize n, nent = snap->nent;
+ /* Store the value of all modified slots to the Lua stack. */
+ for (n = 0; n < nent; n++) {
+ SnapEntry sn = map[n];
+ BCReg s = snap_slot(sn);
+ int32_t ofs = 8*((int32_t)s-1-LJ_FR2);
+ IRRef ref = snap_ref(sn);
+ IRIns *ir = IR(ref);
+ if ((sn & SNAP_NORESTORE))
+ continue;
+ if (irt_isnum(ir->t)) {
+ Reg src = ra_alloc1(as, ref, RSET_FPR);
+ emit_lso(as, A64I_STRd, (src & 31), RID_BASE, ofs);
+ } else {
+ asm_tvstore64(as, RID_BASE, ofs, ref);
+ }
+ checkmclim(as);
+ }
+ lua_assert(map + nent == flinks);
+}
+
+/* -- GC handling --------------------------------------------------------- */
+
+/* Check GC threshold and do one or more GC steps. */
+static void asm_gc_check(ASMState *as)
+{
+ const CCallInfo *ci = &lj_ir_callinfo[IRCALL_lj_gc_step_jit];
+ IRRef args[2];
+ MCLabel l_end;
+ Reg tmp1, tmp2;
+ ra_evictset(as, RSET_SCRATCH);
+ l_end = emit_label(as);
+ /* Exit trace if in GCSatomic or GCSfinalize. Avoids syncing GC objects. */
+ asm_guardcnb(as, A64I_CBNZ, RID_RET); /* Assumes asm_snap_prep() is done. */
+ args[0] = ASMREF_TMP1; /* global_State *g */
+ args[1] = ASMREF_TMP2; /* MSize steps */
+ asm_gencall(as, ci, args);
+ tmp1 = ra_releasetmp(as, ASMREF_TMP1);
+ tmp2 = ra_releasetmp(as, ASMREF_TMP2);
+ emit_loadi(as, tmp2, as->gcsteps);
+ /* Jump around GC step if GC total < GC threshold. */
+ emit_cond_branch(as, CC_LS, l_end);
+ emit_nm(as, A64I_CMPx, RID_TMP, tmp2);
+ emit_lso(as, A64I_LDRx, tmp2, tmp1,
+ (int32_t)offsetof(global_State, gc.threshold));
+ emit_lso(as, A64I_LDRx, RID_TMP, tmp1,
+ (int32_t)offsetof(global_State, gc.total));
+ ra_allockreg(as, i64ptr(J2G(as->J)), tmp1);
+ as->gcsteps = 0;
+ checkmclim(as);
+}
+
+/* -- Loop handling ------------------------------------------------------- */
+
+/* Fixup the loop branch. */
+static void asm_loop_fixup(ASMState *as)
+{
+ MCode *p = as->mctop;
+ MCode *target = as->mcp;
+ if (as->loopinv) { /* Inverted loop branch? */
+ uint32_t mask = (p[-2] & 0x7e000000) == 0x36000000 ? 0x3fffu : 0x7ffffu;
+ ptrdiff_t delta = target - (p - 2);
+ /* asm_guard* already inverted the bcc/tnb/cnb and patched the final b. */
+ p[-2] |= ((uint32_t)delta & mask) << 5;
+ } else {
+ ptrdiff_t delta = target - (p - 1);
+ p[-1] = A64I_B | ((uint32_t)(delta) & 0x03ffffffu);
+ }
+}
+
+/* -- Head of trace ------------------------------------------------------- */
+
+/* Reload L register from g->cur_L. */
+static void asm_head_lreg(ASMState *as)
+{
+ IRIns *ir = IR(ASMREF_L);
+ if (ra_used(ir)) {
+ Reg r = ra_dest(as, ir, RSET_GPR);
+ emit_getgl(as, r, cur_L);
+ ra_evictk(as);
+ }
+}
+
+/* Coalesce BASE register for a root trace. */
+static void asm_head_root_base(ASMState *as)
+{
+ IRIns *ir;
+ asm_head_lreg(as);
+ ir = IR(REF_BASE);
+ if (ra_hasreg(ir->r) && (rset_test(as->modset, ir->r) || irt_ismarked(ir->t)))
+ ra_spill(as, ir);
+ ra_destreg(as, ir, RID_BASE);
+}
+
+/* Coalesce BASE register for a side trace. */
+static RegSet asm_head_side_base(ASMState *as, IRIns *irp, RegSet allow)
+{
+ IRIns *ir;
+ asm_head_lreg(as);
+ ir = IR(REF_BASE);
+ if (ra_hasreg(ir->r) && (rset_test(as->modset, ir->r) || irt_ismarked(ir->t)))
+ ra_spill(as, ir);
+ if (ra_hasspill(irp->s)) {
+ rset_clear(allow, ra_dest(as, ir, allow));
+ } else {
+ Reg r = irp->r;
+ lua_assert(ra_hasreg(r));
+ rset_clear(allow, r);
+ if (r != ir->r && !rset_test(as->freeset, r))
+ ra_restore(as, regcost_ref(as->cost[r]));
+ ra_destreg(as, ir, r);
+ }
+ return allow;
+}
+
+/* -- Tail of trace ------------------------------------------------------- */
+
+/* Fixup the tail code. */
+static void asm_tail_fixup(ASMState *as, TraceNo lnk)
+{
+ MCode *p = as->mctop;
+ MCode *target;
+ /* Undo the sp adjustment in BC_JLOOP when exiting to the interpreter. */
+ int32_t spadj = as->T->spadjust + (lnk ? 0 : sps_scale(SPS_FIXED));
+ if (spadj == 0) {
+ *--p = A64I_LE(A64I_NOP);
+ as->mctop = p;
+ } else {
+ /* Patch stack adjustment. */
+ uint32_t k = emit_isk12(spadj);
+ lua_assert(k);
+ p[-2] = (A64I_ADDx^k) | A64F_D(RID_SP) | A64F_N(RID_SP);
+ }
+ /* Patch exit branch. */
+ target = lnk ? traceref(as->J, lnk)->mcode : (MCode *)lj_vm_exit_interp;
+ p[-1] = A64I_B | (((target-p)+1)&0x03ffffffu);
+}
+
+/* Prepare tail of code. */
+static void asm_tail_prep(ASMState *as)
+{
+ MCode *p = as->mctop - 1; /* Leave room for exit branch. */
+ if (as->loopref) {
+ as->invmcp = as->mcp = p;
+ } else {
+ as->mcp = p-1; /* Leave room for stack pointer adjustment. */
+ as->invmcp = NULL;
+ }
+ *p = 0; /* Prevent load/store merging. */
+}
+
+/* -- Trace setup --------------------------------------------------------- */
+
+/* Ensure there are enough stack slots for call arguments. */
+static Reg asm_setup_call_slots(ASMState *as, IRIns *ir, const CCallInfo *ci)
+{
+ IRRef args[CCI_NARGS_MAX*2];
+ uint32_t i, nargs = CCI_XNARGS(ci);
+ int nslots = 0, ngpr = REGARG_NUMGPR, nfpr = REGARG_NUMFPR;
+ asm_collectargs(as, ir, ci, args);
+ for (i = 0; i < nargs; i++) {
+ if (args[i] && irt_isfp(IR(args[i])->t)) {
+ if (nfpr > 0) nfpr--; else nslots += 2;
+ } else {
+ if (ngpr > 0) ngpr--; else nslots += 2;
+ }
+ }
+ if (nslots > as->evenspill) /* Leave room for args in stack slots. */
+ as->evenspill = nslots;
+ return REGSP_HINT(RID_RET);
+}
+
+static void asm_setup_target(ASMState *as)
+{
+ /* May need extra exit for asm_stack_check on side traces. */
+ asm_exitstub_setup(as, as->T->nsnap + (as->parent ? 1 : 0));
+}
+
+#if LJ_BE
+/* ARM64 instructions are always little-endian. Swap for ARM64BE. */
+static void asm_mcode_fixup(MCode *mcode, MSize size)
+{
+ MCode *pe = (MCode *)((char *)mcode + size);
+ while (mcode < pe) {
+ MCode ins = *mcode;
+ *mcode++ = lj_bswap(ins);
+ }
+}
+#define LJ_TARGET_MCODE_FIXUP 1
+#endif
+
+/* -- Trace patching ------------------------------------------------------ */
+
+/* Patch exit jumps of existing machine code to a new target. */
+void lj_asm_patchexit(jit_State *J, GCtrace *T, ExitNo exitno, MCode *target)
+{
+ MCode *p = T->mcode;
+ MCode *pe = (MCode *)((char *)p + T->szmcode);
+ MCode *cstart = NULL, *cend = p;
+ MCode *mcarea = lj_mcode_patch(J, p, 0);
+ MCode *px = exitstub_trace_addr(T, exitno);
+ for (; p < pe; p++) {
+ /* Look for exitstub branch, replace with branch to target. */
+ MCode ins = A64I_LE(*p);
+ if ((ins & 0xff000000u) == 0x54000000u &&
+ ((ins ^ ((px-p)<<5)) & 0x00ffffe0u) == 0) {
+ /* Patch bcc exitstub. */
+ *p = A64I_LE((ins & 0xff00001fu) | (((target-p)<<5) & 0x00ffffe0u));
+ cend = p+1;
+ if (!cstart) cstart = p;
+ } else if ((ins & 0xfc000000u) == 0x14000000u &&
+ ((ins ^ (px-p)) & 0x03ffffffu) == 0) {
+ /* Patch b exitstub. */
+ *p = A64I_LE((ins & 0xfc000000u) | ((target-p) & 0x03ffffffu));
+ cend = p+1;
+ if (!cstart) cstart = p;
+ } else if ((ins & 0x7e000000u) == 0x34000000u &&
+ ((ins ^ ((px-p)<<5)) & 0x00ffffe0u) == 0) {
+ /* Patch cbz/cbnz exitstub. */
+ *p = A64I_LE((ins & 0xff00001f) | (((target-p)<<5) & 0x00ffffe0u));
+ cend = p+1;
+ if (!cstart) cstart = p;
+ } else if ((ins & 0x7e000000u) == 0x36000000u &&
+ ((ins ^ ((px-p)<<5)) & 0x0007ffe0u) == 0) {
+ /* Patch tbz/tbnz exitstub. */
+ *p = A64I_LE((ins & 0xfff8001fu) | (((target-p)<<5) & 0x0007ffe0u));
+ cend = p+1;
+ if (!cstart) cstart = p;
+ }
+ }
+ lua_assert(cstart != NULL);
+ lj_mcode_sync(cstart, cend);
+ lj_mcode_patch(J, mcarea, 1);
+}
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm64.h b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm64.h
new file mode 100644
index 00000000000..6da4c7d4b4e
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_emit_arm64.h
@@ -0,0 +1,419 @@
+/*
+** ARM64 instruction emitter.
+** Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
+**
+** Contributed by Djordje Kovacevic and Stefan Pejic from RT-RK.com.
+** Sponsored by Cisco Systems, Inc.
+*/
+
+/* -- Constant encoding --------------------------------------------------- */
+
+static uint64_t get_k64val(IRIns *ir)
+{
+ if (ir->o == IR_KINT64) {
+ return ir_kint64(ir)->u64;
+ } else if (ir->o == IR_KGC) {
+ return (uint64_t)ir_kgc(ir);
+ } else if (ir->o == IR_KPTR || ir->o == IR_KKPTR) {
+ return (uint64_t)ir_kptr(ir);
+ } else {
+ lua_assert(ir->o == IR_KINT || ir->o == IR_KNULL);
+ return ir->i; /* Sign-extended. */
+ }
+}
+
+/* Encode constant in K12 format for data processing instructions. */
+static uint32_t emit_isk12(int64_t n)
+{
+ uint64_t k = (n < 0) ? -n : n;
+ uint32_t m = (n < 0) ? 0x40000000 : 0;
+ if (k < 0x1000) {
+ return A64I_K12|m|A64F_U12(k);
+ } else if ((k & 0xfff000) == k) {
+ return A64I_K12|m|0x400000|A64F_U12(k>>12);
+ }
+ return 0;
+}
+
+#define emit_clz64(n) __builtin_clzll(n)
+#define emit_ctz64(n) __builtin_ctzll(n)
+
+/* Encode constant in K13 format for logical data processing instructions. */
+static uint32_t emit_isk13(uint64_t n, int is64)
+{
+ int inv = 0, w = 128, lz, tz;
+ if (n & 1) { n = ~n; w = 64; inv = 1; } /* Avoid wrap-around of ones. */
+ if (!n) return 0; /* Neither all-zero nor all-ones are allowed. */
+ do { /* Find the repeat width. */
+ if (is64 && (uint32_t)(n^(n>>32))) break;
+ n = (uint32_t)n;
+ if (!n) return 0; /* Ditto when passing n=0xffffffff and is64=0. */
+ w = 32; if ((n^(n>>16)) & 0xffff) break;
+ n = n & 0xffff; w = 16; if ((n^(n>>8)) & 0xff) break;
+ n = n & 0xff; w = 8; if ((n^(n>>4)) & 0xf) break;
+ n = n & 0xf; w = 4; if ((n^(n>>2)) & 0x3) break;
+ n = n & 0x3; w = 2;
+ } while (0);
+ lz = emit_clz64(n);
+ tz = emit_ctz64(n);
+ if ((int64_t)(n << lz) >> (lz+tz) != -1ll) return 0; /* Non-contiguous? */
+ if (inv)
+ return A64I_K13 | (((lz-w) & 127) << 16) | (((lz+tz-w-1) & 63) << 10);
+ else
+ return A64I_K13 | ((w-tz) << 16) | (((63-lz-tz-w-w) & 63) << 10);
+}
+
+static uint32_t emit_isfpk64(uint64_t n)
+{
+ uint64_t etop9 = ((n >> 54) & 0x1ff);
+ if ((n << 16) == 0 && (etop9 == 0x100 || etop9 == 0x0ff)) {
+ return (uint32_t)(((n >> 48) & 0x7f) | ((n >> 56) & 0x80));
+ }
+ return ~0u;
+}
+
+/* -- Emit basic instructions --------------------------------------------- */
+
+static void emit_dnma(ASMState *as, A64Ins ai, Reg rd, Reg rn, Reg rm, Reg ra)
+{
+ *--as->mcp = ai | A64F_D(rd) | A64F_N(rn) | A64F_M(rm) | A64F_A(ra);
+}
+
+static void emit_dnm(ASMState *as, A64Ins ai, Reg rd, Reg rn, Reg rm)
+{
+ *--as->mcp = ai | A64F_D(rd) | A64F_N(rn) | A64F_M(rm);
+}
+
+static void emit_dm(ASMState *as, A64Ins ai, Reg rd, Reg rm)
+{
+ *--as->mcp = ai | A64F_D(rd) | A64F_M(rm);
+}
+
+static void emit_dn(ASMState *as, A64Ins ai, Reg rd, Reg rn)
+{
+ *--as->mcp = ai | A64F_D(rd) | A64F_N(rn);
+}
+
+static void emit_nm(ASMState *as, A64Ins ai, Reg rn, Reg rm)
+{
+ *--as->mcp = ai | A64F_N(rn) | A64F_M(rm);
+}
+
+static void emit_d(ASMState *as, A64Ins ai, Reg rd)
+{
+ *--as->mcp = ai | A64F_D(rd);
+}
+
+static void emit_n(ASMState *as, A64Ins ai, Reg rn)
+{
+ *--as->mcp = ai | A64F_N(rn);
+}
+
+static int emit_checkofs(A64Ins ai, int64_t ofs)
+{
+ int scale = (ai >> 30) & 3;
+ if (ofs < 0 || (ofs & ((1<<scale)-1))) {
+ return (ofs >= -256 && ofs <= 255) ? -1 : 0;
+ } else {
+ return (ofs < (4096<<scale)) ? 1 : 0;
+ }
+}
+
+static void emit_lso(ASMState *as, A64Ins ai, Reg rd, Reg rn, int64_t ofs)
+{
+ int ot = emit_checkofs(ai, ofs), sc = (ai >> 30) & 3;
+ lua_assert(ot);
+ /* Combine LDR/STR pairs to LDP/STP. */
+ if ((sc == 2 || sc == 3) &&
+ (!(ai & 0x400000) || rd != rn) &&
+ as->mcp != as->mcloop) {
+ uint32_t prev = *as->mcp & ~A64F_D(31);
+ int ofsm = ofs - (1<<sc), ofsp = ofs + (1<<sc);
+ A64Ins aip;
+ if (prev == (ai | A64F_N(rn) | A64F_U12(ofsm>>sc)) ||
+ prev == ((ai^A64I_LS_U) | A64F_N(rn) | A64F_S9(ofsm&0x1ff))) {
+ aip = (A64F_A(rd) | A64F_D(*as->mcp & 31));
+ } else if (prev == (ai | A64F_N(rn) | A64F_U12(ofsp>>sc)) ||
+ prev == ((ai^A64I_LS_U) | A64F_N(rn) | A64F_S9(ofsp&0x1ff))) {
+ aip = (A64F_D(rd) | A64F_A(*as->mcp & 31));
+ ofsm = ofs;
+ } else {
+ goto nopair;
+ }
+ if (ofsm >= (int)((unsigned int)-64<<sc) && ofsm <= (63<<sc)) {
+ *as->mcp = aip | A64F_N(rn) | ((ofsm >> sc) << 15) |
+ (ai ^ ((ai == A64I_LDRx || ai == A64I_STRx) ? 0x50000000 : 0x90000000));
+ return;
+ }
+ }
+nopair:
+ if (ot == 1)
+ *--as->mcp = ai | A64F_D(rd) | A64F_N(rn) | A64F_U12(ofs >> sc);
+ else
+ *--as->mcp = (ai^A64I_LS_U) | A64F_D(rd) | A64F_N(rn) | A64F_S9(ofs & 0x1ff);
+}
+
+/* -- Emit loads/stores --------------------------------------------------- */
+
+/* Prefer rematerialization of BASE/L from global_State over spills. */
+#define emit_canremat(ref) ((ref) <= ASMREF_L)
+
+/* Try to find an N-step delta relative to other consts with N < lim. */
+static int emit_kdelta(ASMState *as, Reg rd, uint64_t k, int lim)
+{
+ RegSet work = ~as->freeset & RSET_GPR;
+ if (lim <= 1) return 0; /* Can't beat that. */
+ while (work) {
+ Reg r = rset_picktop(work);
+ IRRef ref = regcost_ref(as->cost[r]);
+ lua_assert(r != rd);
+ if (ref < REF_TRUE) {
+ uint64_t kx = ra_iskref(ref) ? (uint64_t)ra_krefk(as, ref) :
+ get_k64val(IR(ref));
+ int64_t delta = (int64_t)(k - kx);
+ if (delta == 0) {
+ emit_dm(as, A64I_MOVx, rd, r);
+ return 1;
+ } else {
+ uint32_t k12 = emit_isk12(delta < 0 ? -delta : delta);
+ if (k12) {
+ emit_dn(as, (delta < 0 ? A64I_SUBx : A64I_ADDx)^k12, rd, r);
+ return 1;
+ }
+ /* Do other ops or multi-step deltas pay off? Probably not.
+ ** E.g. XOR rarely helps with pointer consts.
+ */
+ }
+ }
+ rset_clear(work, r);
+ }
+ return 0; /* Failed. */
+}
+
+static void emit_loadk(ASMState *as, Reg rd, uint64_t u64, int is64)
+{
+ uint32_t k13 = emit_isk13(u64, is64);
+ if (k13) { /* Can the constant be represented as a bitmask immediate? */
+ emit_dn(as, (is64|A64I_ORRw)^k13, rd, RID_ZERO);
+ } else {
+ int i, zeros = 0, ones = 0, neg;
+ if (!is64) u64 = (int64_t)(int32_t)u64; /* Sign-extend. */
+ /* Count homogeneous 16 bit fragments. */
+ for (i = 0; i < 4; i++) {
+ uint64_t frag = (u64 >> i*16) & 0xffff;
+ zeros += (frag == 0);
+ ones += (frag == 0xffff);
+ }
+ neg = ones > zeros; /* Use MOVN if it pays off. */
+ if (!emit_kdelta(as, rd, u64, 4 - (neg ? ones : zeros))) {
+ int shift = 0, lshift = 0;
+ uint64_t n64 = neg ? ~u64 : u64;
+ if (n64 != 0) {
+ /* Find first/last fragment to be filled. */
+ shift = (63-emit_clz64(n64)) & ~15;
+ lshift = emit_ctz64(n64) & ~15;
+ }
+ /* MOVK requires the original value (u64). */
+ while (shift > lshift) {
+ uint32_t u16 = (u64 >> shift) & 0xffff;
+ /* Skip fragments that are correctly filled by MOVN/MOVZ. */
+ if (u16 != (neg ? 0xffff : 0))
+ emit_d(as, is64 | A64I_MOVKw | A64F_U16(u16) | A64F_LSL16(shift), rd);
+ shift -= 16;
+ }
+ /* But MOVN needs an inverted value (n64). */
+ emit_d(as, (neg ? A64I_MOVNx : A64I_MOVZx) |
+ A64F_U16((n64 >> lshift) & 0xffff) | A64F_LSL16(lshift), rd);
+ }
+ }
+}
+
+/* Load a 32 bit constant into a GPR. */
+#define emit_loadi(as, rd, i) emit_loadk(as, rd, i, 0)
+
+/* Load a 64 bit constant into a GPR. */
+#define emit_loadu64(as, rd, i) emit_loadk(as, rd, i, A64I_X)
+
+#define emit_loada(as, r, addr) emit_loadu64(as, (r), (uintptr_t)(addr))
+
+#define glofs(as, k) \
+ ((intptr_t)((uintptr_t)(k) - (uintptr_t)&J2GG(as->J)->g))
+#define mcpofs(as, k) \
+ ((intptr_t)((uintptr_t)(k) - (uintptr_t)(as->mcp - 1)))
+#define checkmcpofs(as, k) \
+ ((((mcpofs(as, k)>>2) + 0x00040000) >> 19) == 0)
+
+static Reg ra_allock(ASMState *as, intptr_t k, RegSet allow);
+
+/* Get/set from constant pointer. */
+static void emit_lsptr(ASMState *as, A64Ins ai, Reg r, void *p)
+{
+ /* First, check if ip + offset is in range. */
+ if ((ai & 0x00400000) && checkmcpofs(as, p)) {
+ emit_d(as, A64I_LDRLx | A64F_S19(mcpofs(as, p)>>2), r);
+ } else {
+ Reg base = RID_GL; /* Next, try GL + offset. */
+ int64_t ofs = glofs(as, p);
+ if (!emit_checkofs(ai, ofs)) { /* Else split up into base reg + offset. */
+ int64_t i64 = i64ptr(p);
+ base = ra_allock(as, (i64 & ~0x7fffull), rset_exclude(RSET_GPR, r));
+ ofs = i64 & 0x7fffull;
+ }
+ emit_lso(as, ai, r, base, ofs);
+ }
+}
+
+/* Load 64 bit IR constant into register. */
+static void emit_loadk64(ASMState *as, Reg r, IRIns *ir)
+{
+ const uint64_t *k = &ir_k64(ir)->u64;
+ int64_t ofs;
+ if (r >= RID_MAX_GPR) {
+ uint32_t fpk = emit_isfpk64(*k);
+ if (fpk != ~0u) {
+ emit_d(as, A64I_FMOV_DI | A64F_FP8(fpk), (r & 31));
+ return;
+ }
+ }
+ ofs = glofs(as, k);
+ if (emit_checkofs(A64I_LDRx, ofs)) {
+ emit_lso(as, r >= RID_MAX_GPR ? A64I_LDRd : A64I_LDRx,
+ (r & 31), RID_GL, ofs);
+ } else {
+ if (r >= RID_MAX_GPR) {
+ emit_dn(as, A64I_FMOV_D_R, (r & 31), RID_TMP);
+ r = RID_TMP;
+ }
+ if (checkmcpofs(as, k))
+ emit_d(as, A64I_LDRLx | A64F_S19(mcpofs(as, k)>>2), r);
+ else
+ emit_loadu64(as, r, *k);
+ }
+}
+
+/* Get/set global_State fields. */
+#define emit_getgl(as, r, field) \
+ emit_lsptr(as, A64I_LDRx, (r), (void *)&J2G(as->J)->field)
+#define emit_setgl(as, r, field) \
+ emit_lsptr(as, A64I_STRx, (r), (void *)&J2G(as->J)->field)
+
+/* Trace number is determined from pc of exit instruction. */
+#define emit_setvmstate(as, i) UNUSED(i)
+
+/* -- Emit control-flow instructions -------------------------------------- */
+
+/* Label for internal jumps. */
+typedef MCode *MCLabel;
+
+/* Return label pointing to current PC. */
+#define emit_label(as) ((as)->mcp)
+
+static void emit_cond_branch(ASMState *as, A64CC cond, MCode *target)
+{
+ MCode *p = --as->mcp;
+ ptrdiff_t delta = target - p;
+ lua_assert(((delta + 0x40000) >> 19) == 0);
+ *p = A64I_BCC | A64F_S19(delta) | cond;
+}
+
+static void emit_branch(ASMState *as, A64Ins ai, MCode *target)
+{
+ MCode *p = --as->mcp;
+ ptrdiff_t delta = target - p;
+ lua_assert(((delta + 0x02000000) >> 26) == 0);
+ *p = ai | ((uint32_t)delta & 0x03ffffffu);
+}
+
+static void emit_tnb(ASMState *as, A64Ins ai, Reg r, uint32_t bit, MCode *target)
+{
+ MCode *p = --as->mcp;
+ ptrdiff_t delta = target - p;
+ lua_assert(bit < 63 && ((delta + 0x2000) >> 14) == 0);
+ if (bit > 31) ai |= A64I_X;
+ *p = ai | A64F_BIT(bit & 31) | A64F_S14((uint32_t)delta & 0x3fffu) | r;
+}
+
+static void emit_cnb(ASMState *as, A64Ins ai, Reg r, MCode *target)
+{
+ MCode *p = --as->mcp;
+ ptrdiff_t delta = target - p;
+ lua_assert(((delta + 0x40000) >> 19) == 0);
+ *p = ai | A64F_S19(delta) | r;
+}
+
+#define emit_jmp(as, target) emit_branch(as, A64I_B, (target))
+
+static void emit_call(ASMState *as, void *target)
+{
+ MCode *p = --as->mcp;
+ ptrdiff_t delta = (char *)target - (char *)p;
+ if ((((delta>>2) + 0x02000000) >> 26) == 0) {
+ *p = A64I_BL | ((uint32_t)(delta>>2) & 0x03ffffffu);
+ } else { /* Target out of range: need indirect call. But don't use R0-R7. */
+ Reg r = ra_allock(as, i64ptr(target),
+ RSET_RANGE(RID_X8, RID_MAX_GPR)-RSET_FIXED);
+ *p = A64I_BLR | A64F_N(r);
+ }
+}
+
+/* -- Emit generic operations --------------------------------------------- */
+
+/* Generic move between two regs. */
+static void emit_movrr(ASMState *as, IRIns *ir, Reg dst, Reg src)
+{
+ if (dst >= RID_MAX_GPR) {
+ emit_dn(as, irt_isnum(ir->t) ? A64I_FMOV_D : A64I_FMOV_S,
+ (dst & 31), (src & 31));
+ return;
+ }
+ if (as->mcp != as->mcloop) { /* Swap early registers for loads/stores. */
+ MCode ins = *as->mcp, swp = (src^dst);
+ if ((ins & 0xbf800000) == 0xb9000000) {
+ if (!((ins ^ (dst << 5)) & 0x000003e0))
+ *as->mcp = ins ^ (swp << 5); /* Swap N in load/store. */
+ if (!(ins & 0x00400000) && !((ins ^ dst) & 0x0000001f))
+ *as->mcp = ins ^ swp; /* Swap D in store. */
+ }
+ }
+ emit_dm(as, A64I_MOVx, dst, src);
+}
+
+/* Generic load of register with base and (small) offset address. */
+static void emit_loadofs(ASMState *as, IRIns *ir, Reg r, Reg base, int32_t ofs)
+{
+ if (r >= RID_MAX_GPR)
+ emit_lso(as, irt_isnum(ir->t) ? A64I_LDRd : A64I_LDRs, (r & 31), base, ofs);
+ else
+ emit_lso(as, irt_is64(ir->t) ? A64I_LDRx : A64I_LDRw, r, base, ofs);
+}
+
+/* Generic store of register with base and (small) offset address. */
+static void emit_storeofs(ASMState *as, IRIns *ir, Reg r, Reg base, int32_t ofs)
+{
+ if (r >= RID_MAX_GPR)
+ emit_lso(as, irt_isnum(ir->t) ? A64I_STRd : A64I_STRs, (r & 31), base, ofs);
+ else
+ emit_lso(as, irt_is64(ir->t) ? A64I_STRx : A64I_STRw, r, base, ofs);
+}
+
+/* Emit an arithmetic operation with a constant operand. */
+static void emit_opk(ASMState *as, A64Ins ai, Reg dest, Reg src,
+ int32_t i, RegSet allow)
+{
+ uint32_t k = emit_isk12(i);
+ if (k)
+ emit_dn(as, ai^k, dest, src);
+ else
+ emit_dnm(as, ai, dest, src, ra_allock(as, i, allow));
+}
+
+/* Add offset to pointer. */
+static void emit_addptr(ASMState *as, Reg r, int32_t ofs)
+{
+ if (ofs)
+ emit_opk(as, ofs < 0 ? A64I_SUBx : A64I_ADDx, r, r,
+ ofs < 0 ? -ofs : ofs, rset_exclude(RSET_GPR, r));
+}
+
+#define emit_spsub(as, ofs) emit_addptr(as, RID_SP, -(ofs))
+
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c b/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
index c046402dbed..77f8326ddbd 100644
--- a/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
+++ b/Build/source/libs/luajit/LuaJIT-src/src/lj_load.c
@@ -166,7 +166,16 @@ LUA_API int lua_dump(lua_State *L, lua_Writer writer, void *data)
return 1;
}
-
+/* -- Luajittex needs this one because it's faster than make it Lua -- */
+LUA_API int RESERVED_lua_dump(lua_State *L, lua_Writer writer, void *data, int strip)
+{
+ cTValue *o = L->top-1;
+ api_check(L, L->top > L->base);
+ if (tvisfunc(o) && isluafunc(funcV(o)))
+ return lj_bcwrite(L, funcproto(funcV(o)), writer, data, strip);
+ else
+ return 1;
+}
/* -- Luajittex needs this one because it overloads loadfile -- */
LUALIB_API int RESERVED_load_aux_JIT(lua_State *L, int status, int envarg)
diff --git a/Build/source/libs/luajit/LuaJIT-src/src/vm_mips64.dasc b/Build/source/libs/luajit/LuaJIT-src/src/vm_mips64.dasc
new file mode 100644
index 00000000000..c06270a0afc
--- /dev/null
+++ b/Build/source/libs/luajit/LuaJIT-src/src/vm_mips64.dasc
@@ -0,0 +1,5062 @@
+|// Low-level VM code for MIPS64 CPUs.
+|// Bytecode interpreter, fast functions and helper functions.
+|// Copyright (C) 2005-2017 Mike Pall. See Copyright Notice in luajit.h
+|//
+|// Contributed by Djordje Kovacevic and Stefan Pejic from RT-RK.com.
+|// Sponsored by Cisco Systems, Inc.
+|
+|.arch mips64
+|.section code_op, code_sub
+|
+|.actionlist build_actionlist
+|.globals GLOB_
+|.globalnames globnames
+|.externnames extnames
+|
+|// Note: The ragged indentation of the instructions is intentional.
+|// The starting columns indicate data dependencies.
+|
+|//-----------------------------------------------------------------------
+|
+|// Fixed register assignments for the interpreter.
+|// Don't use: r0 = 0, r26/r27 = reserved, r28 = gp, r29 = sp, r31 = ra
+|
+|.macro .FPU, a, b
+|.if FPU
+| a, b
+|.endif
+|.endmacro
+|
+|// The following must be C callee-save (but BASE is often refetched).
+|.define BASE, r16 // Base of current Lua stack frame.
+|.define KBASE, r17 // Constants of current Lua function.
+|.define PC, r18 // Next PC.
+|.define DISPATCH, r19 // Opcode dispatch table.
+|.define LREG, r20 // Register holding lua_State (also in SAVE_L).
+|.define MULTRES, r21 // Size of multi-result: (nresults+1)*8.
+|
+|.define JGL, r30 // On-trace: global_State + 32768.
+|
+|// Constants for type-comparisons, stores and conversions. C callee-save.
+|.define TISNIL, r30
+|.define TISNUM, r22
+|.if FPU
+|.define TOBIT, f30 // 2^52 + 2^51.
+|.endif
+|
+|// The following temporaries are not saved across C calls, except for RA.
+|.define RA, r23 // Callee-save.
+|.define RB, r8
+|.define RC, r9
+|.define RD, r10
+|.define INS, r11
+|
+|.define AT, r1 // Assembler temporary.
+|.define TMP0, r12
+|.define TMP1, r13
+|.define TMP2, r14
+|.define TMP3, r15
+|
+|// MIPS n64 calling convention.
+|.define CFUNCADDR, r25
+|.define CARG1, r4
+|.define CARG2, r5
+|.define CARG3, r6
+|.define CARG4, r7
+|.define CARG5, r8
+|.define CARG6, r9
+|.define CARG7, r10
+|.define CARG8, r11
+|
+|.define CRET1, r2
+|.define CRET2, r3
+|
+|.if FPU
+|.define FARG1, f12
+|.define FARG2, f13
+|.define FARG3, f14
+|.define FARG4, f15
+|.define FARG5, f16
+|.define FARG6, f17
+|.define FARG7, f18
+|.define FARG8, f19
+|
+|.define FRET1, f0
+|.define FRET2, f2
+|.endif
+|
+|// Stack layout while in interpreter. Must match with lj_frame.h.
+|.if FPU // MIPS64 hard-float.
+|
+|.define CFRAME_SPACE, 192 // Delta for sp.
+|
+|//----- 16 byte aligned, <-- sp entering interpreter
+|.define SAVE_ERRF, 188(sp) // 32 bit values.
+|.define SAVE_NRES, 184(sp)
+|.define SAVE_CFRAME, 176(sp) // 64 bit values.
+|.define SAVE_L, 168(sp)
+|.define SAVE_PC, 160(sp)
+|//----- 16 byte aligned
+|.define SAVE_GPR_, 80 // .. 80+10*8: 64 bit GPR saves.
+|.define SAVE_FPR_, 16 // .. 16+8*8: 64 bit FPR saves.
+|
+|.else // MIPS64 soft-float
+|
+|.define CFRAME_SPACE, 128 // Delta for sp.
+|
+|//----- 16 byte aligned, <-- sp entering interpreter
+|.define SAVE_ERRF, 124(sp) // 32 bit values.
+|.define SAVE_NRES, 120(sp)
+|.define SAVE_CFRAME, 112(sp) // 64 bit values.
+|.define SAVE_L, 104(sp)
+|.define SAVE_PC, 96(sp)
+|//----- 16 byte aligned
+|.define SAVE_GPR_, 16 // .. 16+10*8: 64 bit GPR saves.
+|
+|.endif
+|
+|.define TMPX, 8(sp) // Unused by interpreter, temp for JIT code.
+|.define TMPD, 0(sp)
+|//----- 16 byte aligned
+|
+|.define TMPD_OFS, 0
+|
+|.define SAVE_MULTRES, TMPD
+|
+|//-----------------------------------------------------------------------
+|
+|.macro saveregs
+| daddiu sp, sp, -CFRAME_SPACE
+| sd ra, SAVE_GPR_+9*8(sp)
+| sd r30, SAVE_GPR_+8*8(sp)
+| .FPU sdc1 f31, SAVE_FPR_+7*8(sp)
+| sd r23, SAVE_GPR_+7*8(sp)
+| .FPU sdc1 f30, SAVE_FPR_+6*8(sp)
+| sd r22, SAVE_GPR_+6*8(sp)
+| .FPU sdc1 f29, SAVE_FPR_+5*8(sp)
+| sd r21, SAVE_GPR_+5*8(sp)
+| .FPU sdc1 f28, SAVE_FPR_+4*8(sp)
+| sd r20, SAVE_GPR_+4*8(sp)
+| .FPU sdc1 f27, SAVE_FPR_+3*8(sp)
+| sd r19, SAVE_GPR_+3*8(sp)
+| .FPU sdc1 f26, SAVE_FPR_+2*8(sp)
+| sd r18, SAVE_GPR_+2*8(sp)
+| .FPU sdc1 f25, SAVE_FPR_+1*8(sp)
+| sd r17, SAVE_GPR_+1*8(sp)
+| .FPU sdc1 f24, SAVE_FPR_+0*8(sp)
+| sd r16, SAVE_GPR_+0*8(sp)
+|.endmacro
+|
+|.macro restoreregs_ret
+| ld ra, SAVE_GPR_+9*8(sp)
+| ld r30, SAVE_GPR_+8*8(sp)
+| ld r23, SAVE_GPR_+7*8(sp)
+| .FPU ldc1 f31, SAVE_FPR_+7*8(sp)
+| ld r22, SAVE_GPR_+6*8(sp)
+| .FPU ldc1 f30, SAVE_FPR_+6*8(sp)
+| ld r21, SAVE_GPR_+5*8(sp)
+| .FPU ldc1 f29, SAVE_FPR_+5*8(sp)
+| ld r20, SAVE_GPR_+4*8(sp)
+| .FPU ldc1 f28, SAVE_FPR_+4*8(sp)
+| ld r19, SAVE_GPR_+3*8(sp)
+| .FPU ldc1 f27, SAVE_FPR_+3*8(sp)
+| ld r18, SAVE_GPR_+2*8(sp)
+| .FPU ldc1 f26, SAVE_FPR_+2*8(sp)
+| ld r17, SAVE_GPR_+1*8(sp)
+| .FPU ldc1 f25, SAVE_FPR_+1*8(sp)
+| ld r16, SAVE_GPR_+0*8(sp)
+| .FPU ldc1 f24, SAVE_FPR_+0*8(sp)
+| jr ra
+| daddiu sp, sp, CFRAME_SPACE
+|.endmacro
+|
+|// Type definitions. Some of these are only used for documentation.
+|.type L, lua_State, LREG
+|.type GL, global_State
+|.type TVALUE, TValue
+|.type GCOBJ, GCobj
+|.type STR, GCstr
+|.type TAB, GCtab
+|.type LFUNC, GCfuncL
+|.type CFUNC, GCfuncC
+|.type PROTO, GCproto
+|.type UPVAL, GCupval
+|.type NODE, Node
+|.type NARGS8, int
+|.type TRACE, GCtrace
+|.type SBUF, SBuf
+|
+|//-----------------------------------------------------------------------
+|
+|// Trap for not-yet-implemented parts.
+|.macro NYI; .long 0xf0f0f0f0; .endmacro
+|
+|// Macros to mark delay slots.
+|.macro ., a; a; .endmacro
+|.macro ., a,b; a,b; .endmacro
+|.macro ., a,b,c; a,b,c; .endmacro
+|.macro ., a,b,c,d; a,b,c,d; .endmacro
+|
+|.define FRAME_PC, -8
+|.define FRAME_FUNC, -16
+|
+|//-----------------------------------------------------------------------
+|
+|// Endian-specific defines.
+|.if ENDIAN_LE
+|.define HI, 4
+|.define LO, 0
+|.define OFS_RD, 2
+|.define OFS_RA, 1
+|.define OFS_OP, 0
+|.else
+|.define HI, 0
+|.define LO, 4
+|.define OFS_RD, 0
+|.define OFS_RA, 2
+|.define OFS_OP, 3
+|.endif
+|
+|// Instruction decode.
+|.macro decode_OP1, dst, ins; andi dst, ins, 0xff; .endmacro
+|.macro decode_OP8a, dst, ins; andi dst, ins, 0xff; .endmacro
+|.macro decode_OP8b, dst; sll dst, dst, 3; .endmacro
+|.macro decode_RC8a, dst, ins; srl dst, ins, 13; .endmacro
+|.macro decode_RC8b, dst; andi dst, dst, 0x7f8; .endmacro
+|.macro decode_RD4b, dst; sll dst, dst, 2; .endmacro
+|.macro decode_RA8a, dst, ins; srl dst, ins, 5; .endmacro
+|.macro decode_RA8b, dst; andi dst, dst, 0x7f8; .endmacro
+|.macro decode_RB8a, dst, ins; srl dst, ins, 21; .endmacro
+|.macro decode_RB8b, dst; andi dst, dst, 0x7f8; .endmacro
+|.macro decode_RD8a, dst, ins; srl dst, ins, 16; .endmacro
+|.macro decode_RD8b, dst; sll dst, dst, 3; .endmacro
+|.macro decode_RDtoRC8, dst, src; andi dst, src, 0x7f8; .endmacro
+|
+|// Instruction fetch.
+|.macro ins_NEXT1
+| lw INS, 0(PC)
+| daddiu PC, PC, 4
+|.endmacro
+|// Instruction decode+dispatch.
+|.macro ins_NEXT2
+| decode_OP8a TMP1, INS
+| decode_OP8b TMP1
+| daddu TMP0, DISPATCH, TMP1
+| decode_RD8a RD, INS
+| ld AT, 0(TMP0)
+| decode_RA8a RA, INS
+| decode_RD8b RD
+| jr AT
+| decode_RA8b RA
+|.endmacro
+|.macro ins_NEXT
+| ins_NEXT1
+| ins_NEXT2
+|.endmacro
+|
+|// Instruction footer.
+|.if 1
+| // Replicated dispatch. Less unpredictable branches, but higher I-Cache use.
+| .define ins_next, ins_NEXT
+| .define ins_next_, ins_NEXT
+| .define ins_next1, ins_NEXT1
+| .define ins_next2, ins_NEXT2
+|.else
+| // Common dispatch. Lower I-Cache use, only one (very) unpredictable branch.
+| // Affects only certain kinds of benchmarks (and only with -j off).
+| .macro ins_next
+| b ->ins_next
+| .endmacro
+| .macro ins_next1
+| .endmacro
+| .macro ins_next2
+| b ->ins_next
+| .endmacro
+| .macro ins_next_
+| ->ins_next:
+| ins_NEXT
+| .endmacro
+|.endif
+|
+|// Call decode and dispatch.
+|.macro ins_callt
+| // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, FRAME_PC(BASE) = PC
+| ld PC, LFUNC:RB->pc
+| lw INS, 0(PC)
+| daddiu PC, PC, 4
+| decode_OP8a TMP1, INS
+| decode_RA8a RA, INS
+| decode_OP8b TMP1
+| decode_RA8b RA
+| daddu TMP0, DISPATCH, TMP1
+| ld TMP0, 0(TMP0)
+| jr TMP0
+| daddu RA, RA, BASE
+|.endmacro
+|
+|.macro ins_call
+| // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, PC = caller PC
+| sd PC, FRAME_PC(BASE)
+| ins_callt
+|.endmacro
+|
+|//-----------------------------------------------------------------------
+|
+|.macro branch_RD
+| srl TMP0, RD, 1
+| lui AT, (-(BCBIAS_J*4 >> 16) & 65535)
+| addu TMP0, TMP0, AT
+| daddu PC, PC, TMP0
+|.endmacro
+|
+|// Assumes DISPATCH is relative to GL.
+#define DISPATCH_GL(field) (GG_DISP2G + (int)offsetof(global_State, field))
+#define DISPATCH_J(field) (GG_DISP2J + (int)offsetof(jit_State, field))
+#define GG_DISP2GOT (GG_OFS(got) - GG_OFS(dispatch))
+#define DISPATCH_GOT(name) (GG_DISP2GOT + sizeof(void*)*LJ_GOT_##name)
+|
+#define PC2PROTO(field) ((int)offsetof(GCproto, field)-(int)sizeof(GCproto))
+|
+|.macro load_got, func
+| ld CFUNCADDR, DISPATCH_GOT(func)(DISPATCH)
+|.endmacro
+|// Much faster. Sadly, there's no easy way to force the required code layout.
+|// .macro call_intern, func; bal extern func; .endmacro
+|.macro call_intern, func; jalr CFUNCADDR; .endmacro
+|.macro call_extern; jalr CFUNCADDR; .endmacro
+|.macro jmp_extern; jr CFUNCADDR; .endmacro
+|
+|.macro hotcheck, delta, target
+| dsrl TMP1, PC, 1
+| andi TMP1, TMP1, 126
+| daddu TMP1, TMP1, DISPATCH
+| lhu TMP2, GG_DISP2HOT(TMP1)
+| addiu TMP2, TMP2, -delta
+| bltz TMP2, target
+|. sh TMP2, GG_DISP2HOT(TMP1)
+|.endmacro
+|
+|.macro hotloop
+| hotcheck HOTCOUNT_LOOP, ->vm_hotloop
+|.endmacro
+|
+|.macro hotcall
+| hotcheck HOTCOUNT_CALL, ->vm_hotcall
+|.endmacro
+|
+|// Set current VM state. Uses TMP0.
+|.macro li_vmstate, st; li TMP0, ~LJ_VMST_..st; .endmacro
+|.macro st_vmstate; sw TMP0, DISPATCH_GL(vmstate)(DISPATCH); .endmacro
+|
+|// Move table write barrier back. Overwrites mark and tmp.
+|.macro barrierback, tab, mark, tmp, target
+| ld tmp, DISPATCH_GL(gc.grayagain)(DISPATCH)
+| andi mark, mark, ~LJ_GC_BLACK & 255 // black2gray(tab)
+| sd tab, DISPATCH_GL(gc.grayagain)(DISPATCH)
+| sb mark, tab->marked
+| b target
+|. sd tmp, tab->gclist
+|.endmacro
+|
+|// Clear type tag. Isolate lowest 14+32+1=47 bits of reg.
+|.macro cleartp, reg; dextm reg, reg, 0, 14; .endmacro
+|.macro cleartp, dst, reg; dextm dst, reg, 0, 14; .endmacro
+|
+|// Set type tag: Merge 17 type bits into bits [15+32=47, 31+32+1=64) of dst.
+|.macro settp, dst, tp; dinsu dst, tp, 15, 31; .endmacro
+|
+|// Extract (negative) type tag.
+|.macro gettp, dst, src; dsra dst, src, 47; .endmacro
+|
+|// Macros to check the TValue type and extract the GCobj. Branch on failure.
+|.macro checktp, reg, tp, target
+| gettp AT, reg
+| daddiu AT, AT, tp
+| bnez AT, target
+|. cleartp reg
+|.endmacro
+|.macro checktp, dst, reg, tp, target
+| gettp AT, reg
+| daddiu AT, AT, tp
+| bnez AT, target
+|. cleartp dst, reg
+|.endmacro
+|.macro checkstr, reg, target; checktp reg, -LJ_TSTR, target; .endmacro
+|.macro checktab, reg, target; checktp reg, -LJ_TTAB, target; .endmacro
+|.macro checkfunc, reg, target; checktp reg, -LJ_TFUNC, target; .endmacro
+|.macro checkint, reg, target // Caveat: has delay slot!
+| gettp AT, reg
+| bne AT, TISNUM, target
+|.endmacro
+|.macro checknum, reg, target // Caveat: has delay slot!
+| gettp AT, reg
+| sltiu AT, AT, LJ_TISNUM
+| beqz AT, target
+|.endmacro
+|
+|.macro mov_false, reg
+| lu reg, 0x8000
+| dsll reg, reg, 32
+| not reg, reg
+|.endmacro
+|.macro mov_true, reg
+| li reg, 0x0001
+| dsll reg, reg, 48
+| not reg, reg
+|.endmacro
+|
+|//-----------------------------------------------------------------------
+
+/* Generate subroutines used by opcodes and other parts of the VM. */
+/* The .code_sub section should be last to help static branch prediction. */
+static void build_subroutines(BuildCtx *ctx)
+{
+ |.code_sub
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Return handling ----------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_returnp:
+ | // See vm_return. Also: TMP2 = previous base.
+ | andi AT, PC, FRAME_P
+ | beqz AT, ->cont_dispatch
+ |
+ | // Return from pcall or xpcall fast func.
+ |. mov_true TMP1
+ | ld PC, FRAME_PC(TMP2) // Fetch PC of previous frame.
+ | move BASE, TMP2 // Restore caller base.
+ | // Prepending may overwrite the pcall frame, so do it at the end.
+ | sd TMP1, -8(RA) // Prepend true to results.
+ | daddiu RA, RA, -8
+ |
+ |->vm_returnc:
+ | addiu RD, RD, 8 // RD = (nresults+1)*8.
+ | andi TMP0, PC, FRAME_TYPE
+ | beqz RD, ->vm_unwind_c_eh
+ |. li CRET1, LUA_YIELD
+ | beqz TMP0, ->BC_RET_Z // Handle regular return to Lua.
+ |. move MULTRES, RD
+ |
+ |->vm_return:
+ | // BASE = base, RA = resultptr, RD/MULTRES = (nresults+1)*8, PC = return
+ | // TMP0 = PC & FRAME_TYPE
+ | li TMP2, -8
+ | xori AT, TMP0, FRAME_C
+ | and TMP2, PC, TMP2
+ | bnez AT, ->vm_returnp
+ | dsubu TMP2, BASE, TMP2 // TMP2 = previous base.
+ |
+ | addiu TMP1, RD, -8
+ | sd TMP2, L->base
+ | li_vmstate C
+ | lw TMP2, SAVE_NRES
+ | daddiu BASE, BASE, -16
+ | st_vmstate
+ | beqz TMP1, >2
+ |. sll TMP2, TMP2, 3
+ |1:
+ | addiu TMP1, TMP1, -8
+ | ld CRET1, 0(RA)
+ | daddiu RA, RA, 8
+ | sd CRET1, 0(BASE)
+ | bnez TMP1, <1
+ |. daddiu BASE, BASE, 8
+ |
+ |2:
+ | bne TMP2, RD, >6
+ |3:
+ |. sd BASE, L->top // Store new top.
+ |
+ |->vm_leave_cp:
+ | ld TMP0, SAVE_CFRAME // Restore previous C frame.
+ | move CRET1, r0 // Ok return status for vm_pcall.
+ | sd TMP0, L->cframe
+ |
+ |->vm_leave_unw:
+ | restoreregs_ret
+ |
+ |6:
+ | ld TMP1, L->maxstack
+ | slt AT, TMP2, RD
+ | bnez AT, >7 // Less results wanted?
+ | // More results wanted. Check stack size and fill up results with nil.
+ |. slt AT, BASE, TMP1
+ | beqz AT, >8
+ |. nop
+ | sd TISNIL, 0(BASE)
+ | addiu RD, RD, 8
+ | b <2
+ |. daddiu BASE, BASE, 8
+ |
+ |7: // Less results wanted.
+ | subu TMP0, RD, TMP2
+ | dsubu TMP0, BASE, TMP0 // Either keep top or shrink it.
+ | b <3
+ |. movn BASE, TMP0, TMP2 // LUA_MULTRET+1 case?
+ |
+ |8: // Corner case: need to grow stack for filling up results.
+ | // This can happen if:
+ | // - A C function grows the stack (a lot).
+ | // - The GC shrinks the stack in between.
+ | // - A return back from a lua_call() with (high) nresults adjustment.
+ | load_got lj_state_growstack
+ | move MULTRES, RD
+ | srl CARG2, TMP2, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | lw TMP2, SAVE_NRES
+ | ld BASE, L->top // Need the (realloced) L->top in BASE.
+ | move RD, MULTRES
+ | b <2
+ |. sll TMP2, TMP2, 3
+ |
+ |->vm_unwind_c: // Unwind C stack, return from vm_pcall.
+ | // (void *cframe, int errcode)
+ | move sp, CARG1
+ | move CRET1, CARG2
+ |->vm_unwind_c_eh: // Landing pad for external unwinder.
+ | ld L, SAVE_L
+ | li TMP0, ~LJ_VMST_C
+ | ld GL:TMP1, L->glref
+ | b ->vm_leave_unw
+ |. sw TMP0, GL:TMP1->vmstate
+ |
+ |->vm_unwind_ff: // Unwind C stack, return from ff pcall.
+ | // (void *cframe)
+ | li AT, -4
+ | and sp, CARG1, AT
+ |->vm_unwind_ff_eh: // Landing pad for external unwinder.
+ | ld L, SAVE_L
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | li TISNIL, LJ_TNIL
+ | li TISNUM, LJ_TISNUM
+ | ld BASE, L->base
+ | ld DISPATCH, L->glref // Setup pointer to dispatch table.
+ | .FPU mtc1 TMP3, TOBIT
+ | mov_false TMP1
+ | li_vmstate INTERP
+ | ld PC, FRAME_PC(BASE) // Fetch PC of previous frame.
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | daddiu RA, BASE, -8 // Results start at BASE-8.
+ | daddiu DISPATCH, DISPATCH, GG_G2DISP
+ | sd TMP1, 0(RA) // Prepend false to error message.
+ | st_vmstate
+ | b ->vm_returnc
+ |. li RD, 16 // 2 results: false + error message.
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Grow stack for calls -----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_growstack_c: // Grow stack for C function.
+ | b >2
+ |. li CARG2, LUA_MINSTACK
+ |
+ |->vm_growstack_l: // Grow stack for Lua function.
+ | // BASE = new base, RA = BASE+framesize*8, RC = nargs*8, PC = first PC
+ | daddu RC, BASE, RC
+ | dsubu RA, RA, BASE
+ | sd BASE, L->base
+ | daddiu PC, PC, 4 // Must point after first instruction.
+ | sd RC, L->top
+ | srl CARG2, RA, 3
+ |2:
+ | // L->base = new base, L->top = top
+ | load_got lj_state_growstack
+ | sd PC, SAVE_PC
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | ld BASE, L->base
+ | ld RC, L->top
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | dsubu RC, RC, BASE
+ | cleartp LFUNC:RB
+ | // BASE = new base, RB = LFUNC/CFUNC, RC = nargs*8, FRAME_PC(BASE) = PC
+ | ins_callt // Just retry the call.
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Entry points into the assembler VM ---------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_resume: // Setup C frame and resume thread.
+ | // (lua_State *L, TValue *base, int nres1 = 0, ptrdiff_t ef = 0)
+ | saveregs
+ | move L, CARG1
+ | ld DISPATCH, L->glref // Setup pointer to dispatch table.
+ | move BASE, CARG2
+ | lbu TMP1, L->status
+ | sd L, SAVE_L
+ | li PC, FRAME_CP
+ | daddiu TMP0, sp, CFRAME_RESUME
+ | daddiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw r0, SAVE_NRES
+ | sw r0, SAVE_ERRF
+ | sd CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | sd r0, SAVE_CFRAME
+ | beqz TMP1, >3
+ |. sd TMP0, L->cframe
+ |
+ | // Resume after yield (like a return).
+ | sd L, DISPATCH_GL(cur_L)(DISPATCH)
+ | move RA, BASE
+ | ld BASE, L->base
+ | ld TMP1, L->top
+ | ld PC, FRAME_PC(BASE)
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | dsubu RD, TMP1, BASE
+ | .FPU mtc1 TMP3, TOBIT
+ | sb r0, L->status
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | li_vmstate INTERP
+ | daddiu RD, RD, 8
+ | st_vmstate
+ | move MULTRES, RD
+ | andi TMP0, PC, FRAME_TYPE
+ | li TISNIL, LJ_TNIL
+ | beqz TMP0, ->BC_RET_Z
+ |. li TISNUM, LJ_TISNUM
+ | b ->vm_return
+ |. nop
+ |
+ |->vm_pcall: // Setup protected C frame and enter VM.
+ | // (lua_State *L, TValue *base, int nres1, ptrdiff_t ef)
+ | saveregs
+ | sw CARG4, SAVE_ERRF
+ | b >1
+ |. li PC, FRAME_CP
+ |
+ |->vm_call: // Setup C frame and enter VM.
+ | // (lua_State *L, TValue *base, int nres1)
+ | saveregs
+ | li PC, FRAME_C
+ |
+ |1: // Entry point for vm_pcall above (PC = ftype).
+ | ld TMP1, L:CARG1->cframe
+ | move L, CARG1
+ | sw CARG3, SAVE_NRES
+ | ld DISPATCH, L->glref // Setup pointer to dispatch table.
+ | sd CARG1, SAVE_L
+ | move BASE, CARG2
+ | daddiu DISPATCH, DISPATCH, GG_G2DISP
+ | sd CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | sd TMP1, SAVE_CFRAME
+ | sd sp, L->cframe // Add our C frame to cframe chain.
+ |
+ |3: // Entry point for vm_cpcall/vm_resume (BASE = base, PC = ftype).
+ | sd L, DISPATCH_GL(cur_L)(DISPATCH)
+ | ld TMP2, L->base // TMP2 = old base (used in vmeta_call).
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | ld TMP1, L->top
+ | .FPU mtc1 TMP3, TOBIT
+ | daddu PC, PC, BASE
+ | dsubu NARGS8:RC, TMP1, BASE
+ | li TISNUM, LJ_TISNUM
+ | dsubu PC, PC, TMP2 // PC = frame delta + frame type
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | li_vmstate INTERP
+ | li TISNIL, LJ_TNIL
+ | st_vmstate
+ |
+ |->vm_call_dispatch:
+ | // TMP2 = old base, BASE = new base, RC = nargs*8, PC = caller PC
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | checkfunc LFUNC:RB, ->vmeta_call
+ |
+ |->vm_call_dispatch_f:
+ | ins_call
+ | // BASE = new base, RB = func, RC = nargs*8, PC = caller PC
+ |
+ |->vm_cpcall: // Setup protected C frame, call C.
+ | // (lua_State *L, lua_CFunction func, void *ud, lua_CPFunction cp)
+ | saveregs
+ | move L, CARG1
+ | ld TMP0, L:CARG1->stack
+ | sd CARG1, SAVE_L
+ | ld TMP1, L->top
+ | ld DISPATCH, L->glref // Setup pointer to dispatch table.
+ | sd CARG1, SAVE_PC // Any value outside of bytecode is ok.
+ | dsubu TMP0, TMP0, TMP1 // Compute -savestack(L, L->top).
+ | ld TMP1, L->cframe
+ | daddiu DISPATCH, DISPATCH, GG_G2DISP
+ | sw TMP0, SAVE_NRES // Neg. delta means cframe w/o frame.
+ | sw r0, SAVE_ERRF // No error function.
+ | sd TMP1, SAVE_CFRAME
+ | sd sp, L->cframe // Add our C frame to cframe chain.
+ | sd L, DISPATCH_GL(cur_L)(DISPATCH)
+ | jalr CARG4 // (lua_State *L, lua_CFunction func, void *ud)
+ |. move CFUNCADDR, CARG4
+ | move BASE, CRET1
+ | bnez CRET1, <3 // Else continue with the call.
+ |. li PC, FRAME_CP
+ | b ->vm_leave_cp // No base? Just remove C frame.
+ |. nop
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Metamethod handling ------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// The lj_meta_* functions (except for lj_meta_cat) don't reallocate the
+ |// stack, so BASE doesn't need to be reloaded across these calls.
+ |
+ |//-- Continuation dispatch ----------------------------------------------
+ |
+ |->cont_dispatch:
+ | // BASE = meta base, RA = resultptr, RD = (nresults+1)*8
+ | ld TMP0, -32(BASE) // Continuation.
+ | move RB, BASE
+ | move BASE, TMP2 // Restore caller BASE.
+ | ld LFUNC:TMP1, FRAME_FUNC(TMP2)
+ |.if FFI
+ | sltiu AT, TMP0, 2
+ |.endif
+ | ld PC, -24(RB) // Restore PC from [cont|PC].
+ | cleartp LFUNC:TMP1
+ | daddu TMP2, RA, RD
+ | ld TMP1, LFUNC:TMP1->pc
+ |.if FFI
+ | bnez AT, >1
+ |.endif
+ |. sd TISNIL, -8(TMP2) // Ensure one valid arg.
+ | // BASE = base, RA = resultptr, RB = meta base
+ | jr TMP0 // Jump to continuation.
+ |. ld KBASE, PC2PROTO(k)(TMP1)
+ |
+ |.if FFI
+ |1:
+ | bnez TMP0, ->cont_ffi_callback // cont = 1: return from FFI callback.
+ | // cont = 0: tailcall from C function.
+ |. daddiu TMP1, RB, -32
+ | b ->vm_call_tail
+ |. dsubu RC, TMP1, BASE
+ |.endif
+ |
+ |->cont_cat: // RA = resultptr, RB = meta base
+ | lw INS, -4(PC)
+ | daddiu CARG2, RB, -32
+ | ld CRET1, 0(RA)
+ | decode_RB8a MULTRES, INS
+ | decode_RA8a RA, INS
+ | decode_RB8b MULTRES
+ | decode_RA8b RA
+ | daddu TMP1, BASE, MULTRES
+ | sd BASE, L->base
+ | dsubu CARG3, CARG2, TMP1
+ | bne TMP1, CARG2, ->BC_CAT_Z
+ |. sd CRET1, 0(CARG2)
+ | daddu RA, BASE, RA
+ | b ->cont_nop
+ |. sd CRET1, 0(RA)
+ |
+ |//-- Table indexing metamethods -----------------------------------------
+ |
+ |->vmeta_tgets1:
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TSTR
+ | settp STR:RC, TMP0
+ | b >1
+ |. sd STR:RC, 0(CARG3)
+ |
+ |->vmeta_tgets:
+ | daddiu CARG2, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TTAB
+ | li TMP1, LJ_TSTR
+ | settp TAB:RB, TMP0
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv2)
+ | sd TAB:RB, 0(CARG2)
+ | settp STR:RC, TMP1
+ | b >1
+ |. sd STR:RC, 0(CARG3)
+ |
+ |->vmeta_tgetb: // TMP0 = index
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | settp TMP0, TISNUM
+ | sd TMP0, 0(CARG3)
+ |
+ |->vmeta_tgetv:
+ |1:
+ | load_got lj_meta_tget
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_tget // (lua_State *L, TValue *o, TValue *k)
+ |. move CARG1, L
+ | // Returns TValue * (finished) or NULL (metamethod).
+ | beqz CRET1, >3
+ |. daddiu TMP1, BASE, -FRAME_CONT
+ | ld CARG1, 0(CRET1)
+ | ins_next1
+ | sd CARG1, 0(RA)
+ | ins_next2
+ |
+ |3: // Call __index metamethod.
+ | // BASE = base, L->top = new base, stack = cont/func/t/k
+ | ld BASE, L->top
+ | sd PC, -24(BASE) // [cont|PC]
+ | dsubu PC, BASE, TMP1
+ | ld LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | cleartp LFUNC:RB
+ | b ->vm_call_dispatch_f
+ |. li NARGS8:RC, 16 // 2 args for func(t, k).
+ |
+ |->vmeta_tgetr:
+ | load_got lj_tab_getinth
+ | call_intern lj_tab_getinth // (GCtab *t, int32_t key)
+ |. nop
+ | // Returns cTValue * or NULL.
+ | beqz CRET1, ->BC_TGETR_Z
+ |. move CARG2, TISNIL
+ | b ->BC_TGETR_Z
+ |. ld CARG2, 0(CRET1)
+ |
+ |//-----------------------------------------------------------------------
+ |
+ |->vmeta_tsets1:
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TSTR
+ | settp STR:RC, TMP0
+ | b >1
+ |. sd STR:RC, 0(CARG3)
+ |
+ |->vmeta_tsets:
+ | daddiu CARG2, DISPATCH, DISPATCH_GL(tmptv)
+ | li TMP0, LJ_TTAB
+ | li TMP1, LJ_TSTR
+ | settp TAB:RB, TMP0
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv2)
+ | sd TAB:RB, 0(CARG2)
+ | settp STR:RC, TMP1
+ | b >1
+ |. sd STR:RC, 0(CARG3)
+ |
+ |->vmeta_tsetb: // TMP0 = index
+ | daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | settp TMP0, TISNUM
+ | sd TMP0, 0(CARG3)
+ |
+ |->vmeta_tsetv:
+ |1:
+ | load_got lj_meta_tset
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_tset // (lua_State *L, TValue *o, TValue *k)
+ |. move CARG1, L
+ | // Returns TValue * (finished) or NULL (metamethod).
+ | beqz CRET1, >3
+ |. ld CARG1, 0(RA)
+ | // NOBARRIER: lj_meta_tset ensures the table is not black.
+ | ins_next1
+ | sd CARG1, 0(CRET1)
+ | ins_next2
+ |
+ |3: // Call __newindex metamethod.
+ | // BASE = base, L->top = new base, stack = cont/func/t/k/(v)
+ | daddiu TMP1, BASE, -FRAME_CONT
+ | ld BASE, L->top
+ | sd PC, -24(BASE) // [cont|PC]
+ | dsubu PC, BASE, TMP1
+ | ld LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | cleartp LFUNC:RB
+ | sd CARG1, 16(BASE) // Copy value to third argument.
+ | b ->vm_call_dispatch_f
+ |. li NARGS8:RC, 24 // 3 args for func(t, k, v)
+ |
+ |->vmeta_tsetr:
+ | load_got lj_tab_setinth
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_tab_setinth // (lua_State *L, GCtab *t, int32_t key)
+ |. move CARG1, L
+ | // Returns TValue *.
+ | b ->BC_TSETR_Z
+ |. nop
+ |
+ |//-- Comparison metamethods ---------------------------------------------
+ |
+ |->vmeta_comp:
+ | // RA/RD point to o1/o2.
+ | move CARG2, RA
+ | move CARG3, RD
+ | load_got lj_meta_comp
+ | daddiu PC, PC, -4
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | decode_OP1 CARG4, INS
+ | call_intern lj_meta_comp // (lua_State *L, TValue *o1, *o2, int op)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ |3:
+ | sltiu AT, CRET1, 2
+ | beqz AT, ->vmeta_binop
+ | negu TMP2, CRET1
+ |4:
+ | lhu RD, OFS_RD(PC)
+ | daddiu PC, PC, 4
+ | lui TMP1, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sll RD, RD, 2
+ | addu RD, RD, TMP1
+ | and RD, RD, TMP2
+ | daddu PC, PC, RD
+ |->cont_nop:
+ | ins_next
+ |
+ |->cont_ra: // RA = resultptr
+ | lbu TMP1, -4+OFS_RA(PC)
+ | ld CRET1, 0(RA)
+ | sll TMP1, TMP1, 3
+ | daddu TMP1, BASE, TMP1
+ | b ->cont_nop
+ |. sd CRET1, 0(TMP1)
+ |
+ |->cont_condt: // RA = resultptr
+ | ld TMP0, 0(RA)
+ | gettp TMP0, TMP0
+ | sltiu AT, TMP0, LJ_TISTRUECOND
+ | b <4
+ |. negu TMP2, AT // Branch if result is true.
+ |
+ |->cont_condf: // RA = resultptr
+ | ld TMP0, 0(RA)
+ | gettp TMP0, TMP0
+ | sltiu AT, TMP0, LJ_TISTRUECOND
+ | b <4
+ |. addiu TMP2, AT, -1 // Branch if result is false.
+ |
+ |->vmeta_equal:
+ | // CARG1/CARG2 point to o1/o2. TMP0 is set to 0/1.
+ | load_got lj_meta_equal
+ | cleartp LFUNC:CARG3, CARG2
+ | cleartp LFUNC:CARG2, CARG1
+ | move CARG4, TMP0
+ | daddiu PC, PC, -4
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_equal // (lua_State *L, GCobj *o1, *o2, int ne)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ | b <3
+ |. nop
+ |
+ |->vmeta_equal_cd:
+ |.if FFI
+ | load_got lj_meta_equal_cd
+ | move CARG2, INS
+ | daddiu PC, PC, -4
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_equal_cd // (lua_State *L, BCIns op)
+ |. move CARG1, L
+ | // Returns 0/1 or TValue * (metamethod).
+ | b <3
+ |. nop
+ |.endif
+ |
+ |->vmeta_istype:
+ | load_got lj_meta_istype
+ | daddiu PC, PC, -4
+ | sd BASE, L->base
+ | srl CARG2, RA, 3
+ | srl CARG3, RD, 3
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_istype // (lua_State *L, BCReg ra, BCReg tp)
+ |. move CARG1, L
+ | b ->cont_nop
+ |. nop
+ |
+ |//-- Arithmetic metamethods ---------------------------------------------
+ |
+ |->vmeta_unm:
+ | move RC, RB
+ |
+ |->vmeta_arith:
+ | load_got lj_meta_arith
+ | sd BASE, L->base
+ | move CARG2, RA
+ | sd PC, SAVE_PC
+ | move CARG3, RB
+ | move CARG4, RC
+ | decode_OP1 CARG5, INS // CARG5 == RB.
+ | call_intern lj_meta_arith // (lua_State *L, TValue *ra,*rb,*rc, BCReg op)
+ |. move CARG1, L
+ | // Returns NULL (finished) or TValue * (metamethod).
+ | beqz CRET1, ->cont_nop
+ |. nop
+ |
+ | // Call metamethod for binary op.
+ |->vmeta_binop:
+ | // BASE = old base, CRET1 = new base, stack = cont/func/o1/o2
+ | dsubu TMP1, CRET1, BASE
+ | sd PC, -24(CRET1) // [cont|PC]
+ | move TMP2, BASE
+ | daddiu PC, TMP1, FRAME_CONT
+ | move BASE, CRET1
+ | b ->vm_call_dispatch
+ |. li NARGS8:RC, 16 // 2 args for func(o1, o2).
+ |
+ |->vmeta_len:
+ | // CARG2 already set by BC_LEN.
+#if LJ_52
+ | move MULTRES, CARG1
+#endif
+ | load_got lj_meta_len
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_len // (lua_State *L, TValue *o)
+ |. move CARG1, L
+ | // Returns NULL (retry) or TValue * (metamethod base).
+#if LJ_52
+ | bnez CRET1, ->vmeta_binop // Binop call for compatibility.
+ |. nop
+ | b ->BC_LEN_Z
+ |. move CARG1, MULTRES
+#else
+ | b ->vmeta_binop // Binop call for compatibility.
+ |. nop
+#endif
+ |
+ |//-- Call metamethod ----------------------------------------------------
+ |
+ |->vmeta_call: // Resolve and call __call metamethod.
+ | // TMP2 = old base, BASE = new base, RC = nargs*8
+ | load_got lj_meta_call
+ | sd TMP2, L->base // This is the callers base!
+ | daddiu CARG2, BASE, -16
+ | sd PC, SAVE_PC
+ | daddu CARG3, BASE, RC
+ | move MULTRES, NARGS8:RC
+ | call_intern lj_meta_call // (lua_State *L, TValue *func, TValue *top)
+ |. move CARG1, L
+ | ld LFUNC:RB, FRAME_FUNC(BASE) // Guaranteed to be a function here.
+ | daddiu NARGS8:RC, MULTRES, 8 // Got one more argument now.
+ | cleartp LFUNC:RB
+ | ins_call
+ |
+ |->vmeta_callt: // Resolve __call for BC_CALLT.
+ | // BASE = old base, RA = new base, RC = nargs*8
+ | load_got lj_meta_call
+ | sd BASE, L->base
+ | daddiu CARG2, RA, -16
+ | sd PC, SAVE_PC
+ | daddu CARG3, RA, RC
+ | move MULTRES, NARGS8:RC
+ | call_intern lj_meta_call // (lua_State *L, TValue *func, TValue *top)
+ |. move CARG1, L
+ | ld RB, FRAME_FUNC(RA) // Guaranteed to be a function here.
+ | ld TMP1, FRAME_PC(BASE)
+ | daddiu NARGS8:RC, MULTRES, 8 // Got one more argument now.
+ | b ->BC_CALLT_Z
+ |. cleartp LFUNC:CARG3, RB
+ |
+ |//-- Argument coercion for 'for' statement ------------------------------
+ |
+ |->vmeta_for:
+ | load_got lj_meta_for
+ | sd BASE, L->base
+ | move CARG2, RA
+ | sd PC, SAVE_PC
+ | move MULTRES, INS
+ | call_intern lj_meta_for // (lua_State *L, TValue *base)
+ |. move CARG1, L
+ |.if JIT
+ | decode_OP1 TMP0, MULTRES
+ | li AT, BC_JFORI
+ |.endif
+ | decode_RA8a RA, MULTRES
+ | decode_RD8a RD, MULTRES
+ | decode_RA8b RA
+ |.if JIT
+ | beq TMP0, AT, =>BC_JFORI
+ |. decode_RD8b RD
+ | b =>BC_FORI
+ |. nop
+ |.else
+ | b =>BC_FORI
+ |. decode_RD8b RD
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Fast functions -----------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |.macro .ffunc, name
+ |->ff_ .. name:
+ |.endmacro
+ |
+ |.macro .ffunc_1, name
+ |->ff_ .. name:
+ | beqz NARGS8:RC, ->fff_fallback
+ |. ld CARG1, 0(BASE)
+ |.endmacro
+ |
+ |.macro .ffunc_2, name
+ |->ff_ .. name:
+ | sltiu AT, NARGS8:RC, 16
+ | ld CARG1, 0(BASE)
+ | bnez AT, ->fff_fallback
+ |. ld CARG2, 8(BASE)
+ |.endmacro
+ |
+ |.macro .ffunc_n, name // Caveat: has delay slot!
+ |->ff_ .. name:
+ | ld CARG1, 0(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ | // Either ldc1 or the 1st instruction of checknum is in the delay slot.
+ | .FPU ldc1 FARG1, 0(BASE)
+ | checknum CARG1, ->fff_fallback
+ |.endmacro
+ |
+ |.macro .ffunc_nn, name // Caveat: has delay slot!
+ |->ff_ .. name:
+ | ld CARG1, 0(BASE)
+ | sltiu AT, NARGS8:RC, 16
+ | ld CARG2, 8(BASE)
+ | bnez AT, ->fff_fallback
+ |. gettp TMP0, CARG1
+ | gettp TMP1, CARG2
+ | sltiu TMP0, TMP0, LJ_TISNUM
+ | sltiu TMP1, TMP1, LJ_TISNUM
+ | .FPU ldc1 FARG1, 0(BASE)
+ | and TMP0, TMP0, TMP1
+ | .FPU ldc1 FARG2, 8(BASE)
+ | beqz TMP0, ->fff_fallback
+ |.endmacro
+ |
+ |// Inlined GC threshold check. Caveat: uses TMP0 and TMP1 and has delay slot!
+ |.macro ffgccheck
+ | ld TMP0, DISPATCH_GL(gc.total)(DISPATCH)
+ | ld TMP1, DISPATCH_GL(gc.threshold)(DISPATCH)
+ | dsubu AT, TMP0, TMP1
+ | bgezal AT, ->fff_gcstep
+ |.endmacro
+ |
+ |//-- Base library: checks -----------------------------------------------
+ |.ffunc_1 assert
+ | gettp AT, CARG1
+ | sltiu AT, AT, LJ_TISTRUECOND
+ | beqz AT, ->fff_fallback
+ |. daddiu RA, BASE, -16
+ | ld PC, FRAME_PC(BASE)
+ | addiu RD, NARGS8:RC, 8 // Compute (nresults+1)*8.
+ | daddu TMP2, RA, RD
+ | daddiu TMP1, BASE, 8
+ | beq BASE, TMP2, ->fff_res // Done if exactly 1 argument.
+ |. sd CARG1, 0(RA)
+ |1:
+ | ld CRET1, 0(TMP1)
+ | sd CRET1, -16(TMP1)
+ | bne TMP1, TMP2, <1
+ |. daddiu TMP1, TMP1, 8
+ | b ->fff_res
+ |. nop
+ |
+ |.ffunc_1 type
+ | gettp TMP0, CARG1
+ | sltu TMP1, TISNUM, TMP0
+ | not TMP2, TMP0
+ | li TMP3, ~LJ_TISNUM
+ | movz TMP2, TMP3, TMP1
+ | dsll TMP2, TMP2, 3
+ | daddu TMP2, CFUNC:RB, TMP2
+ | b ->fff_restv
+ |. ld CARG1, CFUNC:TMP2->upvalue
+ |
+ |//-- Base library: getters and setters ---------------------------------
+ |
+ |.ffunc_1 getmetatable
+ | gettp TMP2, CARG1
+ | daddiu TMP0, TMP2, -LJ_TTAB
+ | daddiu TMP1, TMP2, -LJ_TUDATA
+ | movn TMP0, TMP1, TMP0
+ | bnez TMP0, >6
+ |. cleartp TAB:CARG1
+ |1: // Field metatable must be at same offset for GCtab and GCudata!
+ | ld TAB:RB, TAB:CARG1->metatable
+ |2:
+ | ld STR:RC, DISPATCH_GL(gcroot[GCROOT_MMNAME+MM_metatable])(DISPATCH)
+ | beqz TAB:RB, ->fff_restv
+ |. li CARG1, LJ_TNIL
+ | lw TMP0, TAB:RB->hmask
+ | lw TMP1, STR:RC->hash
+ | ld NODE:TMP2, TAB:RB->node
+ | and TMP1, TMP1, TMP0 // idx = str->hash & tab->hmask
+ | dsll TMP0, TMP1, 5
+ | dsll TMP1, TMP1, 3
+ | dsubu TMP1, TMP0, TMP1
+ | daddu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ | li CARG4, LJ_TSTR
+ | settp STR:RC, CARG4 // Tagged key to look for.
+ |3: // Rearranged logic, because we expect _not_ to find the key.
+ | ld TMP0, NODE:TMP2->key
+ | ld CARG1, NODE:TMP2->val
+ | ld NODE:TMP2, NODE:TMP2->next
+ | beq RC, TMP0, >5
+ |. li AT, LJ_TTAB
+ | bnez NODE:TMP2, <3
+ |. nop
+ |4:
+ | move CARG1, RB
+ | b ->fff_restv // Not found, keep default result.
+ |. settp CARG1, AT
+ |5:
+ | bne CARG1, TISNIL, ->fff_restv
+ |. nop
+ | b <4 // Ditto for nil value.
+ |. nop
+ |
+ |6:
+ | sltiu AT, TMP2, LJ_TISNUM
+ | movn TMP2, TISNUM, AT
+ | dsll TMP2, TMP2, 3
+ | dsubu TMP0, DISPATCH, TMP2
+ | b <2
+ |. ld TAB:RB, DISPATCH_GL(gcroot[GCROOT_BASEMT])-8(TMP0)
+ |
+ |.ffunc_2 setmetatable
+ | // Fast path: no mt for table yet and not clearing the mt.
+ | checktp TMP1, CARG1, -LJ_TTAB, ->fff_fallback
+ | gettp TMP3, CARG2
+ | ld TAB:TMP0, TAB:TMP1->metatable
+ | lbu TMP2, TAB:TMP1->marked
+ | daddiu AT, TMP3, -LJ_TTAB
+ | cleartp TAB:CARG2
+ | or AT, AT, TAB:TMP0
+ | bnez AT, ->fff_fallback
+ |. andi AT, TMP2, LJ_GC_BLACK // isblack(table)
+ | beqz AT, ->fff_restv
+ |. sd TAB:CARG2, TAB:TMP1->metatable
+ | barrierback TAB:TMP1, TMP2, TMP0, ->fff_restv
+ |
+ |.ffunc rawget
+ | ld CARG2, 0(BASE)
+ | sltiu AT, NARGS8:RC, 16
+ | load_got lj_tab_get
+ | gettp TMP0, CARG2
+ | cleartp CARG2
+ | daddiu TMP0, TMP0, -LJ_TTAB
+ | or AT, AT, TMP0
+ | bnez AT, ->fff_fallback
+ |. daddiu CARG3, BASE, 8
+ | call_intern lj_tab_get // (lua_State *L, GCtab *t, cTValue *key)
+ |. move CARG1, L
+ | b ->fff_restv
+ |. ld CARG1, 0(CRET1)
+ |
+ |//-- Base library: conversions ------------------------------------------
+ |
+ |.ffunc tonumber
+ | // Only handles the number case inline (without a base argument).
+ | ld CARG1, 0(BASE)
+ | xori AT, NARGS8:RC, 8 // Exactly one number argument.
+ | gettp TMP1, CARG1
+ | sltu TMP0, TISNUM, TMP1
+ | or AT, AT, TMP0
+ | bnez AT, ->fff_fallback
+ |. nop
+ | b ->fff_restv
+ |. nop
+ |
+ |.ffunc_1 tostring
+ | // Only handles the string or number case inline.
+ | gettp TMP0, CARG1
+ | daddiu AT, TMP0, -LJ_TSTR
+ | // A __tostring method in the string base metatable is ignored.
+ | beqz AT, ->fff_restv // String key?
+ | // Handle numbers inline, unless a number base metatable is present.
+ |. ld TMP1, DISPATCH_GL(gcroot[GCROOT_BASEMT_NUM])(DISPATCH)
+ | sltu TMP0, TISNUM, TMP0
+ | or TMP0, TMP0, TMP1
+ | bnez TMP0, ->fff_fallback
+ |. sd BASE, L->base // Add frame since C call can throw.
+ | ffgccheck
+ |. sd PC, SAVE_PC // Redundant (but a defined value).
+ | load_got lj_strfmt_number
+ | move CARG1, L
+ | call_intern lj_strfmt_number // (lua_State *L, cTValue *o)
+ |. move CARG2, BASE
+ | // Returns GCstr *.
+ | li AT, LJ_TSTR
+ | settp CRET1, AT
+ | b ->fff_restv
+ |. move CARG1, CRET1
+ |
+ |//-- Base library: iterators -------------------------------------------
+ |
+ |.ffunc_1 next
+ | checktp CARG2, CARG1, -LJ_TTAB, ->fff_fallback
+ | daddu TMP2, BASE, NARGS8:RC
+ | sd TISNIL, 0(TMP2) // Set missing 2nd arg to nil.
+ | ld PC, FRAME_PC(BASE)
+ | load_got lj_tab_next
+ | sd BASE, L->base // Add frame since C call can throw.
+ | sd BASE, L->top // Dummy frame length is ok.
+ | daddiu CARG3, BASE, 8
+ | sd PC, SAVE_PC
+ | call_intern lj_tab_next // (lua_State *L, GCtab *t, TValue *key)
+ |. move CARG1, L
+ | // Returns 0 at end of traversal.
+ | beqz CRET1, ->fff_restv // End of traversal: return nil.
+ |. move CARG1, TISNIL
+ | ld TMP0, 8(BASE)
+ | daddiu RA, BASE, -16
+ | ld TMP2, 16(BASE)
+ | sd TMP0, 0(RA)
+ | sd TMP2, 8(RA)
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |.ffunc_1 pairs
+ | checktp TAB:TMP1, CARG1, -LJ_TTAB, ->fff_fallback
+ | ld PC, FRAME_PC(BASE)
+#if LJ_52
+ | ld TAB:TMP2, TAB:TMP1->metatable
+ | ld TMP0, CFUNC:RB->upvalue[0]
+ | bnez TAB:TMP2, ->fff_fallback
+#else
+ | ld TMP0, CFUNC:RB->upvalue[0]
+#endif
+ |. daddiu RA, BASE, -16
+ | sd TISNIL, 0(BASE)
+ | sd CARG1, -8(BASE)
+ | sd TMP0, 0(RA)
+ | b ->fff_res
+ |. li RD, (3+1)*8
+ |
+ |.ffunc_2 ipairs_aux
+ | checktab CARG1, ->fff_fallback
+ | checkint CARG2, ->fff_fallback
+ |. lw TMP0, TAB:CARG1->asize
+ | ld TMP1, TAB:CARG1->array
+ | ld PC, FRAME_PC(BASE)
+ | sextw TMP2, CARG2
+ | addiu TMP2, TMP2, 1
+ | sltu AT, TMP2, TMP0
+ | daddiu RA, BASE, -16
+ | zextw TMP0, TMP2
+ | settp TMP0, TISNUM
+ | beqz AT, >2 // Not in array part?
+ |. sd TMP0, 0(RA)
+ | dsll TMP3, TMP2, 3
+ | daddu TMP3, TMP1, TMP3
+ | ld TMP1, 0(TMP3)
+ |1:
+ | beq TMP1, TISNIL, ->fff_res // End of iteration, return 0 results.
+ |. li RD, (0+1)*8
+ | sd TMP1, -8(BASE)
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |2: // Check for empty hash part first. Otherwise call C function.
+ | lw TMP0, TAB:CARG1->hmask
+ | load_got lj_tab_getinth
+ | beqz TMP0, ->fff_res
+ |. li RD, (0+1)*8
+ | call_intern lj_tab_getinth // (GCtab *t, int32_t key)
+ |. move CARG2, TMP2
+ | // Returns cTValue * or NULL.
+ | beqz CRET1, ->fff_res
+ |. li RD, (0+1)*8
+ | b <1
+ |. ld TMP1, 0(CRET1)
+ |
+ |.ffunc_1 ipairs
+ | checktp TAB:TMP1, CARG1, -LJ_TTAB, ->fff_fallback
+ | ld PC, FRAME_PC(BASE)
+#if LJ_52
+ | ld TAB:TMP2, TAB:TMP1->metatable
+ | ld CFUNC:TMP0, CFUNC:RB->upvalue[0]
+ | bnez TAB:TMP2, ->fff_fallback
+#else
+ | ld TMP0, CFUNC:RB->upvalue[0]
+#endif
+ | daddiu RA, BASE, -16
+ | dsll AT, TISNUM, 47
+ | sd CARG1, -8(BASE)
+ | sd AT, 0(BASE)
+ | sd CFUNC:TMP0, 0(RA)
+ | b ->fff_res
+ |. li RD, (3+1)*8
+ |
+ |//-- Base library: catch errors ----------------------------------------
+ |
+ |.ffunc pcall
+ | daddiu NARGS8:RC, NARGS8:RC, -8
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | bltz NARGS8:RC, ->fff_fallback
+ |. move TMP2, BASE
+ | daddiu BASE, BASE, 16
+ | // Remember active hook before pcall.
+ | srl TMP3, TMP3, HOOK_ACTIVE_SHIFT
+ | andi TMP3, TMP3, 1
+ | daddiu PC, TMP3, 16+FRAME_PCALL
+ | beqz NARGS8:RC, ->vm_call_dispatch
+ |1:
+ |. daddu TMP0, BASE, NARGS8:RC
+ |2:
+ | ld TMP1, -16(TMP0)
+ | sd TMP1, -8(TMP0)
+ | daddiu TMP0, TMP0, -8
+ | bne TMP0, BASE, <2
+ |. nop
+ | b ->vm_call_dispatch
+ |. nop
+ |
+ |.ffunc xpcall
+ | daddiu NARGS8:RC, NARGS8:RC, -16
+ | ld CARG1, 0(BASE)
+ | ld CARG2, 8(BASE)
+ | bltz NARGS8:RC, ->fff_fallback
+ |. lbu TMP1, DISPATCH_GL(hookmask)(DISPATCH)
+ | gettp AT, CARG2
+ | daddiu AT, AT, -LJ_TFUNC
+ | bnez AT, ->fff_fallback // Traceback must be a function.
+ |. move TMP2, BASE
+ | daddiu BASE, BASE, 24
+ | // Remember active hook before pcall.
+ | srl TMP3, TMP3, HOOK_ACTIVE_SHIFT
+ | sd CARG2, 0(TMP2) // Swap function and traceback.
+ | andi TMP3, TMP3, 1
+ | sd CARG1, 8(TMP2)
+ | beqz NARGS8:RC, ->vm_call_dispatch
+ |. daddiu PC, TMP3, 24+FRAME_PCALL
+ | b <1
+ |. nop
+ |
+ |//-- Coroutine library --------------------------------------------------
+ |
+ |.macro coroutine_resume_wrap, resume
+ |.if resume
+ |.ffunc_1 coroutine_resume
+ | checktp CARG1, CARG1, -LJ_TTHREAD, ->fff_fallback
+ |.else
+ |.ffunc coroutine_wrap_aux
+ | ld L:CARG1, CFUNC:RB->upvalue[0].gcr
+ | cleartp L:CARG1
+ |.endif
+ | lbu TMP0, L:CARG1->status
+ | ld TMP1, L:CARG1->cframe
+ | ld CARG2, L:CARG1->top
+ | ld TMP2, L:CARG1->base
+ | addiu AT, TMP0, -LUA_YIELD
+ | daddu CARG3, CARG2, TMP0
+ | daddiu TMP3, CARG2, 8
+ | bgtz AT, ->fff_fallback // st > LUA_YIELD?
+ |. movn CARG2, TMP3, AT
+ | xor TMP2, TMP2, CARG3
+ | bnez TMP1, ->fff_fallback // cframe != 0?
+ |. or AT, TMP2, TMP0
+ | ld TMP0, L:CARG1->maxstack
+ | beqz AT, ->fff_fallback // base == top && st == 0?
+ |. ld PC, FRAME_PC(BASE)
+ | daddu TMP2, CARG2, NARGS8:RC
+ | sltu AT, TMP0, TMP2
+ | bnez AT, ->fff_fallback // Stack overflow?
+ |. sd PC, SAVE_PC
+ | sd BASE, L->base
+ |1:
+ |.if resume
+ | daddiu BASE, BASE, 8 // Keep resumed thread in stack for GC.
+ | daddiu NARGS8:RC, NARGS8:RC, -8
+ | daddiu TMP2, TMP2, -8
+ |.endif
+ | sd TMP2, L:CARG1->top
+ | daddu TMP1, BASE, NARGS8:RC
+ | move CARG3, CARG2
+ | sd BASE, L->top
+ |2: // Move args to coroutine.
+ | ld CRET1, 0(BASE)
+ | sltu AT, BASE, TMP1
+ | beqz AT, >3
+ |. daddiu BASE, BASE, 8
+ | sd CRET1, 0(CARG3)
+ | b <2
+ |. daddiu CARG3, CARG3, 8
+ |3:
+ | bal ->vm_resume // (lua_State *L, TValue *base, 0, 0)
+ |. move L:RA, L:CARG1
+ | // Returns thread status.
+ |4:
+ | ld TMP2, L:RA->base
+ | sltiu AT, CRET1, LUA_YIELD+1
+ | ld TMP3, L:RA->top
+ | li_vmstate INTERP
+ | ld BASE, L->base
+ | sd L, DISPATCH_GL(cur_L)(DISPATCH)
+ | st_vmstate
+ | beqz AT, >8
+ |. dsubu RD, TMP3, TMP2
+ | ld TMP0, L->maxstack
+ | beqz RD, >6 // No results?
+ |. daddu TMP1, BASE, RD
+ | sltu AT, TMP0, TMP1
+ | bnez AT, >9 // Need to grow stack?
+ |. daddu TMP3, TMP2, RD
+ | sd TMP2, L:RA->top // Clear coroutine stack.
+ | move TMP1, BASE
+ |5: // Move results from coroutine.
+ | ld CRET1, 0(TMP2)
+ | daddiu TMP2, TMP2, 8
+ | sltu AT, TMP2, TMP3
+ | sd CRET1, 0(TMP1)
+ | bnez AT, <5
+ |. daddiu TMP1, TMP1, 8
+ |6:
+ | andi TMP0, PC, FRAME_TYPE
+ |.if resume
+ | mov_true TMP1
+ | daddiu RA, BASE, -8
+ | sd TMP1, -8(BASE) // Prepend true to results.
+ | daddiu RD, RD, 16
+ |.else
+ | move RA, BASE
+ | daddiu RD, RD, 8
+ |.endif
+ |7:
+ | sd PC, SAVE_PC
+ | beqz TMP0, ->BC_RET_Z
+ |. move MULTRES, RD
+ | b ->vm_return
+ |. nop
+ |
+ |8: // Coroutine returned with error (at co->top-1).
+ |.if resume
+ | daddiu TMP3, TMP3, -8
+ | mov_false TMP1
+ | ld CRET1, 0(TMP3)
+ | sd TMP3, L:RA->top // Remove error from coroutine stack.
+ | li RD, (2+1)*8
+ | sd TMP1, -8(BASE) // Prepend false to results.
+ | daddiu RA, BASE, -8
+ | sd CRET1, 0(BASE) // Copy error message.
+ | b <7
+ |. andi TMP0, PC, FRAME_TYPE
+ |.else
+ | load_got lj_ffh_coroutine_wrap_err
+ | move CARG2, L:RA
+ | call_intern lj_ffh_coroutine_wrap_err // (lua_State *L, lua_State *co)
+ |. move CARG1, L
+ |.endif
+ |
+ |9: // Handle stack expansion on return from yield.
+ | load_got lj_state_growstack
+ | srl CARG2, RD, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | b <4
+ |. li CRET1, 0
+ |.endmacro
+ |
+ | coroutine_resume_wrap 1 // coroutine.resume
+ | coroutine_resume_wrap 0 // coroutine.wrap
+ |
+ |.ffunc coroutine_yield
+ | ld TMP0, L->cframe
+ | daddu TMP1, BASE, NARGS8:RC
+ | sd BASE, L->base
+ | andi TMP0, TMP0, CFRAME_RESUME
+ | sd TMP1, L->top
+ | beqz TMP0, ->fff_fallback
+ |. li CRET1, LUA_YIELD
+ | sd r0, L->cframe
+ | b ->vm_leave_unw
+ |. sb CRET1, L->status
+ |
+ |//-- Math library -------------------------------------------------------
+ |
+ |.ffunc_1 math_abs
+ | gettp CARG2, CARG1
+ | daddiu AT, CARG2, -LJ_TISNUM
+ | bnez AT, >1
+ |. sextw TMP1, CARG1
+ | sra TMP0, TMP1, 31 // Extract sign.
+ | xor TMP1, TMP1, TMP0
+ | dsubu CARG1, TMP1, TMP0
+ | dsll TMP3, CARG1, 32
+ | bgez TMP3, ->fff_restv
+ |. settp CARG1, TISNUM
+ | li CARG1, 0x41e0 // 2^31 as a double.
+ | b ->fff_restv
+ |. dsll CARG1, CARG1, 48
+ |1:
+ | sltiu AT, CARG2, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |. dextm CARG1, CARG1, 0, 30
+ |// fallthrough
+ |
+ |->fff_restv:
+ | // CARG1 = TValue result.
+ | ld PC, FRAME_PC(BASE)
+ | daddiu RA, BASE, -16
+ | sd CARG1, -16(BASE)
+ |->fff_res1:
+ | // RA = results, PC = return.
+ | li RD, (1+1)*8
+ |->fff_res:
+ | // RA = results, RD = (nresults+1)*8, PC = return.
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->vm_return
+ |. move MULTRES, RD
+ | lw INS, -4(PC)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ |5:
+ | sltu AT, RD, RB
+ | bnez AT, >6 // More results expected?
+ |. decode_RA8a TMP0, INS
+ | decode_RA8b TMP0
+ | ins_next1
+ | // Adjust BASE. KBASE is assumed to be set for the calling frame.
+ | dsubu BASE, RA, TMP0
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | daddu TMP1, RA, RD
+ | daddiu RD, RD, 8
+ | b <5
+ |. sd TISNIL, -8(TMP1)
+ |
+ |.macro math_extern, func
+ | .ffunc_n math_ .. func
+ | load_got func
+ | call_extern
+ |. nop
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ |.macro math_extern2, func
+ | .ffunc_nn math_ .. func
+ |. load_got func
+ | call_extern
+ |. nop
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ |// TODO: Return integer type if result is integer (own sf implementation).
+ |.macro math_round, func
+ |->ff_math_ .. func:
+ | ld CARG1, 0(BASE)
+ | beqz NARGS8:RC, ->fff_fallback
+ |. gettp TMP0, CARG1
+ | beq TMP0, TISNUM, ->fff_restv
+ |. sltu AT, TMP0, TISNUM
+ | beqz AT, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ | bal ->vm_ .. func
+ |. nop
+ |.else
+ |. load_got func
+ | call_extern
+ |. nop
+ |.endif
+ | b ->fff_resn
+ |. nop
+ |.endmacro
+ |
+ | math_round floor
+ | math_round ceil
+ |
+ |.ffunc math_log
+ | li AT, 8
+ | bne NARGS8:RC, AT, ->fff_fallback // Exactly 1 argument.
+ |. ld CARG1, 0(BASE)
+ | checknum CARG1, ->fff_fallback
+ |. load_got log
+ |.if FPU
+ | call_extern
+ |. ldc1 FARG1, 0(BASE)
+ |.else
+ | call_extern
+ |. nop
+ |.endif
+ | b ->fff_resn
+ |. nop
+ |
+ | math_extern log10
+ | math_extern exp
+ | math_extern sin
+ | math_extern cos
+ | math_extern tan
+ | math_extern asin
+ | math_extern acos
+ | math_extern atan
+ | math_extern sinh
+ | math_extern cosh
+ | math_extern tanh
+ | math_extern2 pow
+ | math_extern2 atan2
+ | math_extern2 fmod
+ |
+ |.if FPU
+ |.ffunc_n math_sqrt
+ |. sqrt.d FRET1, FARG1
+ |// fallthrough to ->fff_resn
+ |.else
+ | math_extern sqrt
+ |.endif
+ |
+ |->fff_resn:
+ | ld PC, FRAME_PC(BASE)
+ | daddiu RA, BASE, -16
+ | b ->fff_res1
+ |.if FPU
+ |. sdc1 FRET1, 0(RA)
+ |.else
+ |. sd CRET1, 0(RA)
+ |.endif
+ |
+ |
+ |.ffunc_2 math_ldexp
+ | checknum CARG1, ->fff_fallback
+ | checkint CARG2, ->fff_fallback
+ |. load_got ldexp
+ | .FPU ldc1 FARG1, 0(BASE)
+ | call_extern
+ |. lw CARG2, 8+LO(BASE)
+ | b ->fff_resn
+ |. nop
+ |
+ |.ffunc_n math_frexp
+ | load_got frexp
+ | ld PC, FRAME_PC(BASE)
+ | call_extern
+ |. daddiu CARG2, DISPATCH, DISPATCH_GL(tmptv)
+ | lw TMP1, DISPATCH_GL(tmptv)(DISPATCH)
+ | daddiu RA, BASE, -16
+ |.if FPU
+ | mtc1 TMP1, FARG2
+ | sdc1 FRET1, 0(RA)
+ | cvt.d.w FARG2, FARG2
+ | sdc1 FARG2, 8(RA)
+ |.else
+ | sd CRET1, 0(RA)
+ | zextw TMP1, TMP1
+ | settp TMP1, TISNUM
+ | sd TMP1, 8(RA)
+ |.endif
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |.ffunc_n math_modf
+ | load_got modf
+ | ld PC, FRAME_PC(BASE)
+ | call_extern
+ |. daddiu CARG2, BASE, -16
+ | daddiu RA, BASE, -16
+ |.if FPU
+ | sdc1 FRET1, -8(BASE)
+ |.else
+ | sd CRET1, -8(BASE)
+ |.endif
+ | b ->fff_res
+ |. li RD, (2+1)*8
+ |
+ |.macro math_minmax, name, intins, fpins
+ | .ffunc_1 name
+ | daddu TMP3, BASE, NARGS8:RC
+ | checkint CARG1, >5
+ |. daddiu TMP2, BASE, 8
+ |1: // Handle integers.
+ | beq TMP2, TMP3, ->fff_restv
+ |. ld CARG2, 0(TMP2)
+ | checkint CARG2, >3
+ |. sextw CARG1, CARG1
+ | lw CARG2, LO(TMP2)
+ |. slt AT, CARG1, CARG2
+ | intins CARG1, CARG2, AT
+ | daddiu TMP2, TMP2, 8
+ | zextw CARG1, CARG1
+ | b <1
+ |. settp CARG1, TISNUM
+ |
+ |3: // Convert intermediate result to number and continue with number loop.
+ | checknum CARG2, ->fff_fallback
+ |.if FPU
+ |. mtc1 CARG1, FRET1
+ | cvt.d.w FRET1, FRET1
+ | b >7
+ |. ldc1 FARG1, 0(TMP2)
+ |.else
+ |. nop
+ | bal ->vm_sfi2d_1
+ |. nop
+ | b >7
+ |. nop
+ |.endif
+ |
+ |5:
+ | .FPU ldc1 FRET1, 0(BASE)
+ | checknum CARG1, ->fff_fallback
+ |6: // Handle numbers.
+ |. ld CARG2, 0(TMP2)
+ | beq TMP2, TMP3, ->fff_resn
+ |.if FPU
+ | ldc1 FARG1, 0(TMP2)
+ |.else
+ | move CRET1, CARG1
+ |.endif
+ | checknum CARG2, >8
+ |. nop
+ |7:
+ |.if FPU
+ | c.olt.d FRET1, FARG1
+ | fpins FRET1, FARG1
+ |.else
+ | bal ->vm_sfcmpolt
+ |. nop
+ | intins CARG1, CARG2, CRET1
+ |.endif
+ | b <6
+ |. daddiu TMP2, TMP2, 8
+ |
+ |8: // Convert integer to number and continue with number loop.
+ | checkint CARG2, ->fff_fallback
+ |.if FPU
+ |. lwc1 FARG1, LO(TMP2)
+ | b <7
+ |. cvt.d.w FARG1, FARG1
+ |.else
+ |. lw CARG2, LO(TMP2)
+ | bal ->vm_sfi2d_2
+ |. nop
+ | b <7
+ |. nop
+ |.endif
+ |
+ |.endmacro
+ |
+ | math_minmax math_min, movz, movf.d
+ | math_minmax math_max, movn, movt.d
+ |
+ |//-- String library -----------------------------------------------------
+ |
+ |.ffunc string_byte // Only handle the 1-arg case here.
+ | ld CARG1, 0(BASE)
+ | gettp TMP0, CARG1
+ | xori AT, NARGS8:RC, 8
+ | daddiu TMP0, TMP0, -LJ_TSTR
+ | or AT, AT, TMP0
+ | bnez AT, ->fff_fallback // Need exactly 1 string argument.
+ |. cleartp STR:CARG1
+ | lw TMP0, STR:CARG1->len
+ | daddiu RA, BASE, -16
+ | ld PC, FRAME_PC(BASE)
+ | sltu RD, r0, TMP0
+ | lbu TMP1, STR:CARG1[1] // Access is always ok (NUL at end).
+ | addiu RD, RD, 1
+ | sll RD, RD, 3 // RD = ((str->len != 0)+1)*8
+ | settp TMP1, TISNUM
+ | b ->fff_res
+ |. sd TMP1, 0(RA)
+ |
+ |.ffunc string_char // Only handle the 1-arg case here.
+ | ffgccheck
+ |. nop
+ | ld CARG1, 0(BASE)
+ | gettp TMP0, CARG1
+ | xori AT, NARGS8:RC, 8 // Exactly 1 argument.
+ | daddiu TMP0, TMP0, -LJ_TISNUM // Integer.
+ | li TMP1, 255
+ | sextw CARG1, CARG1
+ | or AT, AT, TMP0
+ | sltu TMP1, TMP1, CARG1 // !(255 < n).
+ | or AT, AT, TMP1
+ | bnez AT, ->fff_fallback
+ |. li CARG3, 1
+ | daddiu CARG2, sp, TMPD_OFS
+ | sb CARG1, TMPD
+ |->fff_newstr:
+ | load_got lj_str_new
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | call_intern lj_str_new // (lua_State *L, char *str, size_t l)
+ |. move CARG1, L
+ | // Returns GCstr *.
+ | ld BASE, L->base
+ |->fff_resstr:
+ | li AT, LJ_TSTR
+ | settp CRET1, AT
+ | b ->fff_restv
+ |. move CARG1, CRET1
+ |
+ |.ffunc string_sub
+ | ffgccheck
+ |. nop
+ | addiu AT, NARGS8:RC, -16
+ | ld TMP0, 0(BASE)
+ | bltz AT, ->fff_fallback
+ |. gettp TMP3, TMP0
+ | cleartp STR:CARG1, TMP0
+ | ld CARG2, 8(BASE)
+ | beqz AT, >1
+ |. li CARG4, -1
+ | ld CARG3, 16(BASE)
+ | checkint CARG3, ->fff_fallback
+ |. sextw CARG4, CARG3
+ |1:
+ | checkint CARG2, ->fff_fallback
+ |. li AT, LJ_TSTR
+ | bne TMP3, AT, ->fff_fallback
+ |. sextw CARG3, CARG2
+ | lw CARG2, STR:CARG1->len
+ | // STR:CARG1 = str, CARG2 = str->len, CARG3 = start, CARG4 = end
+ | slt AT, CARG4, r0
+ | addiu TMP0, CARG2, 1
+ | addu TMP1, CARG4, TMP0
+ | slt TMP3, CARG3, r0
+ | movn CARG4, TMP1, AT // if (end < 0) end += len+1
+ | addu TMP1, CARG3, TMP0
+ | movn CARG3, TMP1, TMP3 // if (start < 0) start += len+1
+ | li TMP2, 1
+ | slt AT, CARG4, r0
+ | slt TMP3, r0, CARG3
+ | movn CARG4, r0, AT // if (end < 0) end = 0
+ | movz CARG3, TMP2, TMP3 // if (start < 1) start = 1
+ | slt AT, CARG2, CARG4
+ | movn CARG4, CARG2, AT // if (end > len) end = len
+ | daddu CARG2, STR:CARG1, CARG3
+ | subu CARG3, CARG4, CARG3 // len = end - start
+ | daddiu CARG2, CARG2, sizeof(GCstr)-1
+ | bgez CARG3, ->fff_newstr
+ |. addiu CARG3, CARG3, 1 // len++
+ |->fff_emptystr: // Return empty string.
+ | li AT, LJ_TSTR
+ | daddiu STR:CARG1, DISPATCH, DISPATCH_GL(strempty)
+ | b ->fff_restv
+ |. settp CARG1, AT
+ |
+ |.macro ffstring_op, name
+ | .ffunc string_ .. name
+ | ffgccheck
+ |. nop
+ | beqz NARGS8:RC, ->fff_fallback
+ |. ld CARG2, 0(BASE)
+ | checkstr STR:CARG2, ->fff_fallback
+ | daddiu SBUF:CARG1, DISPATCH, DISPATCH_GL(tmpbuf)
+ | load_got lj_buf_putstr_ .. name
+ | ld TMP0, SBUF:CARG1->b
+ | sd L, SBUF:CARG1->L
+ | sd BASE, L->base
+ | sd TMP0, SBUF:CARG1->p
+ | call_intern extern lj_buf_putstr_ .. name
+ |. sd PC, SAVE_PC
+ | load_got lj_buf_tostr
+ | call_intern lj_buf_tostr
+ |. move SBUF:CARG1, SBUF:CRET1
+ | b ->fff_resstr
+ |. ld BASE, L->base
+ |.endmacro
+ |
+ |ffstring_op reverse
+ |ffstring_op lower
+ |ffstring_op upper
+ |
+ |//-- Bit library --------------------------------------------------------
+ |
+ |->vm_tobit_fb:
+ | beqz TMP1, ->fff_fallback
+ |.if FPU
+ |. ldc1 FARG1, 0(BASE)
+ | add.d FARG1, FARG1, TOBIT
+ | mfc1 CRET1, FARG1
+ | jr ra
+ |. zextw CRET1, CRET1
+ |.else
+ |// FP number to bit conversion for soft-float.
+ |->vm_tobit:
+ | dsll TMP0, CARG1, 1
+ | li CARG3, 1076
+ | dsrl AT, TMP0, 53
+ | dsubu CARG3, CARG3, AT
+ | sltiu AT, CARG3, 54
+ | beqz AT, >1
+ |. dextm TMP0, TMP0, 0, 20
+ | dinsu TMP0, AT, 21, 21
+ | slt AT, CARG1, r0
+ | dsrlv CRET1, TMP0, CARG3
+ | dsubu TMP0, r0, CRET1
+ | movn CRET1, TMP0, AT
+ | jr ra
+ |. zextw CRET1, CRET1
+ |1:
+ | jr ra
+ |. move CRET1, r0
+ |.endif
+ |
+ |.macro .ffunc_bit, name
+ | .ffunc_1 bit_..name
+ | gettp TMP0, CARG1
+ | beq TMP0, TISNUM, >6
+ |. zextw CRET1, CARG1
+ | bal ->vm_tobit_fb
+ |. sltiu TMP1, TMP0, LJ_TISNUM
+ |6:
+ |.endmacro
+ |
+ |.macro .ffunc_bit_op, name, bins
+ | .ffunc_bit name
+ | daddiu TMP2, BASE, 8
+ | daddu TMP3, BASE, NARGS8:RC
+ |1:
+ | beq TMP2, TMP3, ->fff_resi
+ |. ld CARG1, 0(TMP2)
+ | gettp TMP0, CARG1
+ |.if FPU
+ | bne TMP0, TISNUM, >2
+ |. daddiu TMP2, TMP2, 8
+ | zextw CARG1, CARG1
+ | b <1
+ |. bins CRET1, CRET1, CARG1
+ |2:
+ | ldc1 FARG1, -8(TMP2)
+ | sltiu AT, TMP0, LJ_TISNUM
+ | beqz AT, ->fff_fallback
+ |. add.d FARG1, FARG1, TOBIT
+ | mfc1 CARG1, FARG1
+ | zextw CARG1, CARG1
+ | b <1
+ |. bins CRET1, CRET1, CARG1
+ |.else
+ | beq TMP0, TISNUM, >2
+ |. move CRET2, CRET1
+ | bal ->vm_tobit_fb
+ |. sltiu TMP1, TMP0, LJ_TISNUM
+ | move CARG1, CRET2
+ |2:
+ | zextw CARG1, CARG1
+ | bins CRET1, CRET1, CARG1
+ | b <1
+ |. daddiu TMP2, TMP2, 8
+ |.endif
+ |.endmacro
+ |
+ |.ffunc_bit_op band, and
+ |.ffunc_bit_op bor, or
+ |.ffunc_bit_op bxor, xor
+ |
+ |.ffunc_bit bswap
+ | dsrl TMP0, CRET1, 8
+ | dsrl TMP1, CRET1, 24
+ | andi TMP2, TMP0, 0xff00
+ | dins TMP1, CRET1, 24, 31
+ | dins TMP2, TMP0, 16, 23
+ | b ->fff_resi
+ |. or CRET1, TMP1, TMP2
+ |
+ |.ffunc_bit bnot
+ | not CRET1, CRET1
+ | b ->fff_resi
+ |. zextw CRET1, CRET1
+ |
+ |.macro .ffunc_bit_sh, name, shins, shmod
+ | .ffunc_2 bit_..name
+ | gettp TMP0, CARG1
+ | beq TMP0, TISNUM, >1
+ |. nop
+ | bal ->vm_tobit_fb
+ |. sltiu TMP1, TMP0, LJ_TISNUM
+ | move CARG1, CRET1
+ |1:
+ | gettp TMP0, CARG2
+ | bne TMP0, TISNUM, ->fff_fallback
+ |. zextw CARG2, CARG2
+ | sextw CARG1, CARG1
+ |.if shmod == 1
+ | negu CARG2, CARG2
+ |.endif
+ | shins CRET1, CARG1, CARG2
+ | b ->fff_resi
+ |. zextw CRET1, CRET1
+ |.endmacro
+ |
+ |.ffunc_bit_sh lshift, sllv, 0
+ |.ffunc_bit_sh rshift, srlv, 0
+ |.ffunc_bit_sh arshift, srav, 0
+ |.ffunc_bit_sh rol, rotrv, 1
+ |.ffunc_bit_sh ror, rotrv, 0
+ |
+ |.ffunc_bit tobit
+ |->fff_resi:
+ | ld PC, FRAME_PC(BASE)
+ | daddiu RA, BASE, -16
+ | settp CRET1, TISNUM
+ | b ->fff_res1
+ |. sd CRET1, -16(BASE)
+ |
+ |//-----------------------------------------------------------------------
+ |->fff_fallback: // Call fast function fallback handler.
+ | // BASE = new base, RB = CFUNC, RC = nargs*8
+ | ld TMP3, CFUNC:RB->f
+ | daddu TMP1, BASE, NARGS8:RC
+ | ld PC, FRAME_PC(BASE) // Fallback may overwrite PC.
+ | daddiu TMP0, TMP1, 8*LUA_MINSTACK
+ | ld TMP2, L->maxstack
+ | sd PC, SAVE_PC // Redundant (but a defined value).
+ | sltu AT, TMP2, TMP0
+ | sd BASE, L->base
+ | sd TMP1, L->top
+ | bnez AT, >5 // Need to grow stack.
+ |. move CFUNCADDR, TMP3
+ | jalr TMP3 // (lua_State *L)
+ |. move CARG1, L
+ | // Either throws an error, or recovers and returns -1, 0 or nresults+1.
+ | ld BASE, L->base
+ | sll RD, CRET1, 3
+ | bgtz CRET1, ->fff_res // Returned nresults+1?
+ |. daddiu RA, BASE, -16
+ |1: // Returned 0 or -1: retry fast path.
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | ld TMP0, L->top
+ | cleartp LFUNC:RB
+ | bnez CRET1, ->vm_call_tail // Returned -1?
+ |. dsubu NARGS8:RC, TMP0, BASE
+ | ins_callt // Returned 0: retry fast path.
+ |
+ |// Reconstruct previous base for vmeta_call during tailcall.
+ |->vm_call_tail:
+ | andi TMP0, PC, FRAME_TYPE
+ | li AT, -4
+ | bnez TMP0, >3
+ |. and TMP1, PC, AT
+ | lbu TMP1, OFS_RA(PC)
+ | sll TMP1, TMP1, 3
+ | addiu TMP1, TMP1, 16
+ |3:
+ | b ->vm_call_dispatch // Resolve again for tailcall.
+ |. dsubu TMP2, BASE, TMP1
+ |
+ |5: // Grow stack for fallback handler.
+ | load_got lj_state_growstack
+ | li CARG2, LUA_MINSTACK
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | ld BASE, L->base
+ | b <1
+ |. li CRET1, 0 // Force retry.
+ |
+ |->fff_gcstep: // Call GC step function.
+ | // BASE = new base, RC = nargs*8
+ | move MULTRES, ra
+ | load_got lj_gc_step
+ | sd BASE, L->base
+ | daddu TMP0, BASE, NARGS8:RC
+ | sd PC, SAVE_PC // Redundant (but a defined value).
+ | sd TMP0, L->top
+ | call_intern lj_gc_step // (lua_State *L)
+ |. move CARG1, L
+ | ld BASE, L->base
+ | move ra, MULTRES
+ | ld TMP0, L->top
+ | ld CFUNC:RB, FRAME_FUNC(BASE)
+ | cleartp CFUNC:RB
+ | jr ra
+ |. dsubu NARGS8:RC, TMP0, BASE
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Special dispatch targets -------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |->vm_record: // Dispatch target for recording phase.
+ |.if JIT
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | andi AT, TMP3, HOOK_VMEVENT // No recording while in vmevent.
+ | bnez AT, >5
+ | // Decrement the hookcount for consistency, but always do the call.
+ |. lw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE
+ | bnez AT, >1
+ |. addiu TMP2, TMP2, -1
+ | andi AT, TMP3, LUA_MASKLINE|LUA_MASKCOUNT
+ | beqz AT, >1
+ |. nop
+ | b >1
+ |. sw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ |.endif
+ |
+ |->vm_rethook: // Dispatch target for return hooks.
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE // Hook already active?
+ | beqz AT, >1
+ |5: // Re-dispatch to static ins.
+ |. ld AT, GG_DISP2STATIC(TMP0) // Assumes TMP0 holds DISPATCH+OP*4.
+ | jr AT
+ |. nop
+ |
+ |->vm_inshook: // Dispatch target for instr/line hooks.
+ | lbu TMP3, DISPATCH_GL(hookmask)(DISPATCH)
+ | lw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, HOOK_ACTIVE // Hook already active?
+ | bnez AT, <5
+ |. andi AT, TMP3, LUA_MASKLINE|LUA_MASKCOUNT
+ | beqz AT, <5
+ |. addiu TMP2, TMP2, -1
+ | beqz TMP2, >1
+ |. sw TMP2, DISPATCH_GL(hookcount)(DISPATCH)
+ | andi AT, TMP3, LUA_MASKLINE
+ | beqz AT, <5
+ |1:
+ |. load_got lj_dispatch_ins
+ | sw MULTRES, SAVE_MULTRES
+ | move CARG2, PC
+ | sd BASE, L->base
+ | // SAVE_PC must hold the _previous_ PC. The callee updates it with PC.
+ | call_intern lj_dispatch_ins // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ |3:
+ | ld BASE, L->base
+ |4: // Re-dispatch to static ins.
+ | lw INS, -4(PC)
+ | decode_OP8a TMP1, INS
+ | decode_OP8b TMP1
+ | daddu TMP0, DISPATCH, TMP1
+ | decode_RD8a RD, INS
+ | ld AT, GG_DISP2STATIC(TMP0)
+ | decode_RA8a RA, INS
+ | decode_RD8b RD
+ | jr AT
+ | decode_RA8b RA
+ |
+ |->cont_hook: // Continue from hook yield.
+ | daddiu PC, PC, 4
+ | b <4
+ |. lw MULTRES, -24+LO(RB) // Restore MULTRES for *M ins.
+ |
+ |->vm_hotloop: // Hot loop counter underflow.
+ |.if JIT
+ | ld LFUNC:TMP1, FRAME_FUNC(BASE)
+ | daddiu CARG1, DISPATCH, GG_DISP2J
+ | cleartp LFUNC:TMP1
+ | sd PC, SAVE_PC
+ | ld TMP1, LFUNC:TMP1->pc
+ | move CARG2, PC
+ | sd L, DISPATCH_J(L)(DISPATCH)
+ | lbu TMP1, PC2PROTO(framesize)(TMP1)
+ | load_got lj_trace_hot
+ | sd BASE, L->base
+ | dsll TMP1, TMP1, 3
+ | daddu TMP1, BASE, TMP1
+ | call_intern lj_trace_hot // (jit_State *J, const BCIns *pc)
+ |. sd TMP1, L->top
+ | b <3
+ |. nop
+ |.endif
+ |
+ |
+ |->vm_callhook: // Dispatch target for call hooks.
+ |.if JIT
+ | b >1
+ |.endif
+ |. move CARG2, PC
+ |
+ |->vm_hotcall: // Hot call counter underflow.
+ |.if JIT
+ | ori CARG2, PC, 1
+ |1:
+ |.endif
+ | load_got lj_dispatch_call
+ | daddu TMP0, BASE, RC
+ | sd PC, SAVE_PC
+ | sd BASE, L->base
+ | dsubu RA, RA, BASE
+ | sd TMP0, L->top
+ | call_intern lj_dispatch_call // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ | // Returns ASMFunction.
+ | ld BASE, L->base
+ | ld TMP0, L->top
+ | sd r0, SAVE_PC // Invalidate for subsequent line hook.
+ | dsubu NARGS8:RC, TMP0, BASE
+ | daddu RA, BASE, RA
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | cleartp LFUNC:RB
+ | jr CRET1
+ |. lw INS, -4(PC)
+ |
+ |->cont_stitch: // Trace stitching.
+ |.if JIT
+ | // RA = resultptr, RB = meta base
+ | lw INS, -4(PC)
+ | ld TRACE:TMP2, -40(RB) // Save previous trace.
+ | decode_RA8a RC, INS
+ | daddiu AT, MULTRES, -8
+ | cleartp TRACE:TMP2
+ | decode_RA8b RC
+ | beqz AT, >2
+ |. daddu RC, BASE, RC // Call base.
+ |1: // Move results down.
+ | ld CARG1, 0(RA)
+ | daddiu AT, AT, -8
+ | daddiu RA, RA, 8
+ | sd CARG1, 0(RC)
+ | bnez AT, <1
+ |. daddiu RC, RC, 8
+ |2:
+ | decode_RA8a RA, INS
+ | decode_RB8a RB, INS
+ | decode_RA8b RA
+ | decode_RB8b RB
+ | daddu RA, RA, RB
+ | daddu RA, BASE, RA
+ |3:
+ | sltu AT, RC, RA
+ | bnez AT, >9 // More results wanted?
+ |. nop
+ |
+ | lhu TMP3, TRACE:TMP2->traceno
+ | lhu RD, TRACE:TMP2->link
+ | beq RD, TMP3, ->cont_nop // Blacklisted.
+ |. load_got lj_dispatch_stitch
+ | bnez RD, =>BC_JLOOP // Jump to stitched trace.
+ |. sll RD, RD, 3
+ |
+ | // Stitch a new trace to the previous trace.
+ | sw TMP3, DISPATCH_J(exitno)(DISPATCH)
+ | sd L, DISPATCH_J(L)(DISPATCH)
+ | sd BASE, L->base
+ | daddiu CARG1, DISPATCH, GG_DISP2J
+ | call_intern lj_dispatch_stitch // (jit_State *J, const BCIns *pc)
+ |. move CARG2, PC
+ | b ->cont_nop
+ |. ld BASE, L->base
+ |
+ |9:
+ | sd TISNIL, 0(RC)
+ | b <3
+ |. daddiu RC, RC, 8
+ |.endif
+ |
+ |->vm_profhook: // Dispatch target for profiler hook.
+#if LJ_HASPROFILE
+ | load_got lj_dispatch_profile
+ | sw MULTRES, SAVE_MULTRES
+ | move CARG2, PC
+ | sd BASE, L->base
+ | call_intern lj_dispatch_profile // (lua_State *L, const BCIns *pc)
+ |. move CARG1, L
+ | // HOOK_PROFILE is off again, so re-dispatch to dynamic instruction.
+ | daddiu PC, PC, -4
+ | b ->cont_nop
+ |. ld BASE, L->base
+#endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Trace exit handler -------------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |.macro savex_, a, b
+ |.if FPU
+ | sdc1 f..a, a*8(sp)
+ | sdc1 f..b, b*8(sp)
+ | sd r..a, 32*8+a*8(sp)
+ | sd r..b, 32*8+b*8(sp)
+ |.else
+ | sd r..a, a*8(sp)
+ | sd r..b, b*8(sp)
+ |.endif
+ |.endmacro
+ |
+ |->vm_exit_handler:
+ |.if JIT
+ |.if FPU
+ | daddiu sp, sp, -(32*8+32*8)
+ |.else
+ | daddiu sp, sp, -(32*8)
+ |.endif
+ | savex_ 0, 1
+ | savex_ 2, 3
+ | savex_ 4, 5
+ | savex_ 6, 7
+ | savex_ 8, 9
+ | savex_ 10, 11
+ | savex_ 12, 13
+ | savex_ 14, 15
+ | savex_ 16, 17
+ | savex_ 18, 19
+ | savex_ 20, 21
+ | savex_ 22, 23
+ | savex_ 24, 25
+ | savex_ 26, 27
+ | savex_ 28, 30
+ |.if FPU
+ | sdc1 f29, 29*8(sp)
+ | sdc1 f31, 31*8(sp)
+ | sd r0, 32*8+31*8(sp) // Clear RID_TMP.
+ | daddiu TMP2, sp, 32*8+32*8 // Recompute original value of sp.
+ | sd TMP2, 32*8+29*8(sp) // Store sp in RID_SP
+ |.else
+ | sd r0, 31*8(sp) // Clear RID_TMP.
+ | daddiu TMP2, sp, 32*8 // Recompute original value of sp.
+ | sd TMP2, 29*8(sp) // Store sp in RID_SP
+ |.endif
+ | li_vmstate EXIT
+ | daddiu DISPATCH, JGL, -GG_DISP2G-32768
+ | lw TMP1, 0(TMP2) // Load exit number.
+ | st_vmstate
+ | ld L, DISPATCH_GL(cur_L)(DISPATCH)
+ | ld BASE, DISPATCH_GL(jit_base)(DISPATCH)
+ | load_got lj_trace_exit
+ | sd L, DISPATCH_J(L)(DISPATCH)
+ | sw ra, DISPATCH_J(parent)(DISPATCH) // Store trace number.
+ | sd BASE, L->base
+ | sw TMP1, DISPATCH_J(exitno)(DISPATCH) // Store exit number.
+ | daddiu CARG1, DISPATCH, GG_DISP2J
+ | sd r0, DISPATCH_GL(jit_base)(DISPATCH)
+ | call_intern lj_trace_exit // (jit_State *J, ExitState *ex)
+ |. move CARG2, sp
+ | // Returns MULTRES (unscaled) or negated error code.
+ | ld TMP1, L->cframe
+ | li AT, -4
+ | ld BASE, L->base
+ | and sp, TMP1, AT
+ | ld PC, SAVE_PC // Get SAVE_PC.
+ | b >1
+ |. sd L, SAVE_L // Set SAVE_L (on-trace resume/yield).
+ |.endif
+ |->vm_exit_interp:
+ |.if JIT
+ | // CRET1 = MULTRES or negated error code, BASE, PC and JGL set.
+ | ld L, SAVE_L
+ | daddiu DISPATCH, JGL, -GG_DISP2G-32768
+ | sd BASE, L->base
+ |1:
+ | bltz CRET1, >9 // Check for error from exit.
+ |. ld LFUNC:RB, FRAME_FUNC(BASE)
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | dsll MULTRES, CRET1, 3
+ | cleartp LFUNC:RB
+ | sw MULTRES, SAVE_MULTRES
+ | li TISNIL, LJ_TNIL
+ | li TISNUM, LJ_TISNUM // Setup type comparison constants.
+ | .FPU mtc1 TMP3, TOBIT
+ | ld TMP1, LFUNC:RB->pc
+ | sd r0, DISPATCH_GL(jit_base)(DISPATCH)
+ | ld KBASE, PC2PROTO(k)(TMP1)
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | // Modified copy of ins_next which handles function header dispatch, too.
+ | lw INS, 0(PC)
+ | daddiu PC, PC, 4
+ | // Assumes TISNIL == ~LJ_VMST_INTERP == -1
+ | sw TISNIL, DISPATCH_GL(vmstate)(DISPATCH)
+ | decode_OP8a TMP1, INS
+ | decode_OP8b TMP1
+ | sltiu TMP2, TMP1, BC_FUNCF*8
+ | daddu TMP0, DISPATCH, TMP1
+ | decode_RD8a RD, INS
+ | ld AT, 0(TMP0)
+ | decode_RA8a RA, INS
+ | beqz TMP2, >2
+ |. decode_RA8b RA
+ | jr AT
+ |. decode_RD8b RD
+ |2:
+ | sltiu TMP2, TMP1, (BC_FUNCC+2)*8 // Fast function?
+ | bnez TMP2, >3
+ |. ld TMP1, FRAME_PC(BASE)
+ | // Check frame below fast function.
+ | andi TMP0, TMP1, FRAME_TYPE
+ | bnez TMP0, >3 // Trace stitching continuation?
+ |. nop
+ | // Otherwise set KBASE for Lua function below fast function.
+ | lw TMP2, -4(TMP1)
+ | decode_RA8a TMP0, TMP2
+ | decode_RA8b TMP0
+ | dsubu TMP1, BASE, TMP0
+ | ld LFUNC:TMP2, -32(TMP1)
+ | cleartp LFUNC:TMP2
+ | ld TMP1, LFUNC:TMP2->pc
+ | ld KBASE, PC2PROTO(k)(TMP1)
+ |3:
+ | daddiu RC, MULTRES, -8
+ | jr AT
+ |. daddu RA, RA, BASE
+ |
+ |9: // Rethrow error from the right C frame.
+ | load_got lj_err_throw
+ | negu CARG2, CRET1
+ | call_intern lj_err_throw // (lua_State *L, int errcode)
+ |. move CARG1, L
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Math helper functions ----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// Hard-float round to integer.
+ |// Modifies AT, TMP0, FRET1, FRET2, f4. Keeps all others incl. FARG1.
+ |.macro vm_round_hf, func
+ | lui TMP0, 0x4330 // Hiword of 2^52 (double).
+ | dsll TMP0, TMP0, 32
+ | dmtc1 TMP0, f4
+ | abs.d FRET2, FARG1 // |x|
+ | dmfc1 AT, FARG1
+ | c.olt.d 0, FRET2, f4
+ | add.d FRET1, FRET2, f4 // (|x| + 2^52) - 2^52
+ | bc1f 0, >1 // Truncate only if |x| < 2^52.
+ |. sub.d FRET1, FRET1, f4
+ | slt AT, AT, r0
+ |.if "func" == "ceil"
+ | lui TMP0, 0xbff0 // Hiword of -1 (double). Preserves -0.
+ |.else
+ | lui TMP0, 0x3ff0 // Hiword of +1 (double).
+ |.endif
+ |.if "func" == "trunc"
+ | dsll TMP0, TMP0, 32
+ | dmtc1 TMP0, f4
+ | c.olt.d 0, FRET2, FRET1 // |x| < result?
+ | sub.d FRET2, FRET1, f4
+ | movt.d FRET1, FRET2, 0 // If yes, subtract +1.
+ | neg.d FRET2, FRET1
+ | jr ra
+ |. movn.d FRET1, FRET2, AT // Merge sign bit back in.
+ |.else
+ | neg.d FRET2, FRET1
+ | dsll TMP0, TMP0, 32
+ | dmtc1 TMP0, f4
+ | movn.d FRET1, FRET2, AT // Merge sign bit back in.
+ |.if "func" == "ceil"
+ | c.olt.d 0, FRET1, FARG1 // x > result?
+ |.else
+ | c.olt.d 0, FARG1, FRET1 // x < result?
+ |.endif
+ | sub.d FRET2, FRET1, f4 // If yes, subtract +-1.
+ | jr ra
+ |. movt.d FRET1, FRET2, 0
+ |.endif
+ |1:
+ | jr ra
+ |. mov.d FRET1, FARG1
+ |.endmacro
+ |
+ |.macro vm_round, func
+ |.if FPU
+ | vm_round_hf, func
+ |.endif
+ |.endmacro
+ |
+ |->vm_floor:
+ | vm_round floor
+ |->vm_ceil:
+ | vm_round ceil
+ |->vm_trunc:
+ |.if JIT
+ | vm_round trunc
+ |.endif
+ |
+ |// Soft-float integer to number conversion.
+ |.macro sfi2d, ARG
+ |.if not FPU
+ | beqz ARG, >9 // Handle zero first.
+ |. sra TMP0, ARG, 31
+ | xor TMP1, ARG, TMP0
+ | dsubu TMP1, TMP1, TMP0 // Absolute value in TMP1.
+ | dclz ARG, TMP1
+ | addiu ARG, ARG, -11
+ | li AT, 0x3ff+63-11-1
+ | dsllv TMP1, TMP1, ARG // Align mantissa left with leading 1.
+ | subu ARG, AT, ARG // Exponent - 1.
+ | ins ARG, TMP0, 11, 11 // Sign | Exponent.
+ | dsll ARG, ARG, 52 // Align left.
+ | jr ra
+ |. daddu ARG, ARG, TMP1 // Add mantissa, increment exponent.
+ |9:
+ | jr ra
+ |. nop
+ |.endif
+ |.endmacro
+ |
+ |// Input CARG1. Output: CARG1. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfi2d_1:
+ | sfi2d CARG1
+ |
+ |// Input CARG2. Output: CARG2. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfi2d_2:
+ | sfi2d CARG2
+ |
+ |// Soft-float comparison. Equivalent to c.eq.d.
+ |// Input: CARG*. Output: CRET1. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfcmpeq:
+ |.if not FPU
+ | dsll AT, CARG1, 1
+ | dsll TMP0, CARG2, 1
+ | or TMP1, AT, TMP0
+ | beqz TMP1, >8 // Both args +-0: return 1.
+ |. lui TMP1, 0xffe0
+ | dsll TMP1, TMP1, 32
+ | sltu AT, TMP1, AT
+ | sltu TMP0, TMP1, TMP0
+ | or TMP1, AT, TMP0
+ | bnez TMP1, >9 // Either arg is NaN: return 0;
+ |. xor AT, CARG1, CARG2
+ | jr ra
+ |. sltiu CRET1, AT, 1 // Same values: return 1.
+ |8:
+ | jr ra
+ |. li CRET1, 1
+ |9:
+ | jr ra
+ |. li CRET1, 0
+ |.endif
+ |
+ |// Soft-float comparison. Equivalent to c.ult.d and c.olt.d.
+ |// Input: CARG1, CARG2. Output: CRET1. Temporaries: AT, TMP0, TMP1, CRET2.
+ |->vm_sfcmpult:
+ |.if not FPU
+ | b >1
+ |. li CRET2, 1
+ |.endif
+ |
+ |->vm_sfcmpolt:
+ |.if not FPU
+ | li CRET2, 0
+ |1:
+ | dsll AT, CARG1, 1
+ | dsll TMP0, CARG2, 1
+ | or TMP1, AT, TMP0
+ | beqz TMP1, >8 // Both args +-0: return 0.
+ |. lui TMP1, 0xffe0
+ | dsll TMP1, TMP1, 32
+ | sltu AT, TMP1, AT
+ | sltu TMP0, TMP1, TMP0
+ | or TMP1, AT, TMP0
+ | bnez TMP1, >9 // Either arg is NaN: return 0 or 1;
+ |. and AT, CARG1, CARG2
+ | bltz AT, >5 // Both args negative?
+ |. nop
+ | jr ra
+ |. slt CRET1, CARG1, CARG2
+ |5: // Swap conditions if both operands are negative.
+ | jr ra
+ |. slt CRET1, CARG2, CARG1
+ |8:
+ | jr ra
+ |. nop
+ |9:
+ | jr ra
+ |. move CRET1, CRET2
+ |.endif
+ |
+ |// Soft-float comparison. Equivalent to c.ole.d a, b or c.ole.d b, a.
+ |// Input: CARG1, CARG2, TMP3. Output: CRET1. Temporaries: AT, TMP0, TMP1.
+ |->vm_sfcmpolex:
+ |.if not FPU
+ | dsll AT, CARG1, 1
+ | dsll TMP0, CARG2, 1
+ | or TMP1, AT, TMP0
+ | beqz TMP1, >8 // Both args +-0: return 1.
+ |. lui TMP1, 0xffe0
+ | dsll TMP1, TMP1, 32
+ | sltu AT, TMP1, AT
+ | sltu TMP0, TMP1, TMP0
+ | or TMP1, AT, TMP0
+ | bnez TMP1, >9 // Either arg is NaN: return 0;
+ |. and AT, CARG1, CARG2
+ | xor AT, AT, TMP3
+ | bltz AT, >5 // Both args negative?
+ |. nop
+ | jr ra
+ |. slt CRET1, CARG2, CARG1
+ |5: // Swap conditions if both operands are negative.
+ | jr ra
+ |. slt CRET1, CARG1, CARG2
+ |8:
+ | jr ra
+ |. li CRET1, 1
+ |9:
+ | jr ra
+ |. li CRET1, 0
+ |.endif
+ |
+ |//-----------------------------------------------------------------------
+ |//-- Miscellaneous functions --------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |//-----------------------------------------------------------------------
+ |//-- FFI helper functions -----------------------------------------------
+ |//-----------------------------------------------------------------------
+ |
+ |// Handler for callback functions. Callback slot number in r1, g in r2.
+ |->vm_ffi_callback:
+ |.if FFI
+ |.type CTSTATE, CTState, PC
+ | saveregs
+ | ld CTSTATE, GL:r2->ctype_state
+ | daddiu DISPATCH, r2, GG_G2DISP
+ | load_got lj_ccallback_enter
+ | sw r1, CTSTATE->cb.slot
+ | sd CARG1, CTSTATE->cb.gpr[0]
+ | .FPU sdc1 FARG1, CTSTATE->cb.fpr[0]
+ | sd CARG2, CTSTATE->cb.gpr[1]
+ | .FPU sdc1 FARG2, CTSTATE->cb.fpr[1]
+ | sd CARG3, CTSTATE->cb.gpr[2]
+ | .FPU sdc1 FARG3, CTSTATE->cb.fpr[2]
+ | sd CARG4, CTSTATE->cb.gpr[3]
+ | .FPU sdc1 FARG4, CTSTATE->cb.fpr[3]
+ | sd CARG5, CTSTATE->cb.gpr[4]
+ | .FPU sdc1 FARG5, CTSTATE->cb.fpr[4]
+ | sd CARG6, CTSTATE->cb.gpr[5]
+ | .FPU sdc1 FARG6, CTSTATE->cb.fpr[5]
+ | sd CARG7, CTSTATE->cb.gpr[6]
+ | .FPU sdc1 FARG7, CTSTATE->cb.fpr[6]
+ | sd CARG8, CTSTATE->cb.gpr[7]
+ | .FPU sdc1 FARG8, CTSTATE->cb.fpr[7]
+ | daddiu TMP0, sp, CFRAME_SPACE
+ | sd TMP0, CTSTATE->cb.stack
+ | sd r0, SAVE_PC // Any value outside of bytecode is ok.
+ | move CARG2, sp
+ | call_intern lj_ccallback_enter // (CTState *cts, void *cf)
+ |. move CARG1, CTSTATE
+ | // Returns lua_State *.
+ | ld BASE, L:CRET1->base
+ | ld RC, L:CRET1->top
+ | move L, CRET1
+ | .FPU lui TMP3, 0x59c0 // TOBIT = 2^52 + 2^51 (float).
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | .FPU mtc1 TMP3, TOBIT
+ | li TISNIL, LJ_TNIL
+ | li TISNUM, LJ_TISNUM
+ | li_vmstate INTERP
+ | subu RC, RC, BASE
+ | cleartp LFUNC:RB
+ | st_vmstate
+ | .FPU cvt.d.s TOBIT, TOBIT
+ | ins_callt
+ |.endif
+ |
+ |->cont_ffi_callback: // Return from FFI callback.
+ |.if FFI
+ | load_got lj_ccallback_leave
+ | ld CTSTATE, DISPATCH_GL(ctype_state)(DISPATCH)
+ | sd BASE, L->base
+ | sd RB, L->top
+ | sd L, CTSTATE->L
+ | move CARG2, RA
+ | call_intern lj_ccallback_leave // (CTState *cts, TValue *o)
+ |. move CARG1, CTSTATE
+ | .FPU ldc1 FRET1, CTSTATE->cb.fpr[0]
+ | ld CRET1, CTSTATE->cb.gpr[0]
+ | .FPU ldc1 FRET2, CTSTATE->cb.fpr[1]
+ | b ->vm_leave_unw
+ |. ld CRET2, CTSTATE->cb.gpr[1]
+ |.endif
+ |
+ |->vm_ffi_call: // Call C function via FFI.
+ | // Caveat: needs special frame unwinding, see below.
+ |.if FFI
+ | .type CCSTATE, CCallState, CARG1
+ | lw TMP1, CCSTATE->spadj
+ | lbu CARG2, CCSTATE->nsp
+ | move TMP2, sp
+ | dsubu sp, sp, TMP1
+ | sd ra, -8(TMP2)
+ | sll CARG2, CARG2, 3
+ | sd r16, -16(TMP2)
+ | sd CCSTATE, -24(TMP2)
+ | move r16, TMP2
+ | daddiu TMP1, CCSTATE, offsetof(CCallState, stack)
+ | move TMP2, sp
+ | beqz CARG2, >2
+ |. daddu TMP3, TMP1, CARG2
+ |1:
+ | ld TMP0, 0(TMP1)
+ | daddiu TMP1, TMP1, 8
+ | sltu AT, TMP1, TMP3
+ | sd TMP0, 0(TMP2)
+ | bnez AT, <1
+ |. daddiu TMP2, TMP2, 8
+ |2:
+ | ld CFUNCADDR, CCSTATE->func
+ | .FPU ldc1 FARG1, CCSTATE->gpr[0]
+ | ld CARG2, CCSTATE->gpr[1]
+ | .FPU ldc1 FARG2, CCSTATE->gpr[1]
+ | ld CARG3, CCSTATE->gpr[2]
+ | .FPU ldc1 FARG3, CCSTATE->gpr[2]
+ | ld CARG4, CCSTATE->gpr[3]
+ | .FPU ldc1 FARG4, CCSTATE->gpr[3]
+ | ld CARG5, CCSTATE->gpr[4]
+ | .FPU ldc1 FARG5, CCSTATE->gpr[4]
+ | ld CARG6, CCSTATE->gpr[5]
+ | .FPU ldc1 FARG6, CCSTATE->gpr[5]
+ | ld CARG7, CCSTATE->gpr[6]
+ | .FPU ldc1 FARG7, CCSTATE->gpr[6]
+ | ld CARG8, CCSTATE->gpr[7]
+ | .FPU ldc1 FARG8, CCSTATE->gpr[7]
+ | jalr CFUNCADDR
+ |. ld CARG1, CCSTATE->gpr[0] // Do this last, since CCSTATE is CARG1.
+ | ld CCSTATE:TMP1, -24(r16)
+ | ld TMP2, -16(r16)
+ | ld ra, -8(r16)
+ | sd CRET1, CCSTATE:TMP1->gpr[0]
+ | sd CRET2, CCSTATE:TMP1->gpr[1]
+ |.if FPU
+ | sdc1 FRET1, CCSTATE:TMP1->fpr[0]
+ | sdc1 FRET2, CCSTATE:TMP1->fpr[1]
+ |.else
+ | sd CARG1, CCSTATE:TMP1->gpr[2] // 2nd FP struct field for soft-float.
+ |.endif
+ | move sp, r16
+ | jr ra
+ |. move r16, TMP2
+ |.endif
+ |// Note: vm_ffi_call must be the last function in this object file!
+ |
+ |//-----------------------------------------------------------------------
+}
+
+/* Generate the code for a single instruction. */
+static void build_ins(BuildCtx *ctx, BCOp op, int defop)
+{
+ int vk = 0;
+ |=>defop:
+
+ switch (op) {
+
+ /* -- Comparison ops ---------------------------------------------------- */
+
+ /* Remember: all ops branch for a true comparison, fall through otherwise. */
+
+ case BC_ISLT: case BC_ISGE: case BC_ISLE: case BC_ISGT:
+ | // RA = src1*8, RD = src2*8, JMP with RD = target
+ |.macro bc_comp, FRA, FRD, ARGRA, ARGRD, movop, fmovop, fcomp, sfcomp
+ | daddu RA, BASE, RA
+ | daddu RD, BASE, RD
+ | ld ARGRA, 0(RA)
+ | ld ARGRD, 0(RD)
+ | lhu TMP2, OFS_RD(PC)
+ | gettp CARG3, ARGRA
+ | gettp CARG4, ARGRD
+ | bne CARG3, TISNUM, >2
+ |. daddiu PC, PC, 4
+ | bne CARG4, TISNUM, >5
+ |. decode_RD4b TMP2
+ | sextw ARGRA, ARGRA
+ | sextw ARGRD, ARGRD
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | slt AT, CARG1, CARG2
+ | addu TMP2, TMP2, TMP3
+ | movop TMP2, r0, AT
+ |1:
+ | daddu PC, PC, TMP2
+ | ins_next
+ |
+ |2: // RA is not an integer.
+ | sltiu AT, CARG3, LJ_TISNUM
+ | beqz AT, ->vmeta_comp
+ |. lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sltiu AT, CARG4, LJ_TISNUM
+ | beqz AT, >4
+ |. decode_RD4b TMP2
+ |.if FPU
+ | ldc1 FRA, 0(RA)
+ | ldc1 FRD, 0(RD)
+ |.endif
+ |3: // RA and RD are both numbers.
+ |.if FPU
+ | fcomp f20, f22
+ | addu TMP2, TMP2, TMP3
+ | b <1
+ |. fmovop TMP2, r0
+ |.else
+ | bal sfcomp
+ |. addu TMP2, TMP2, TMP3
+ | b <1
+ |. movop TMP2, r0, CRET1
+ |.endif
+ |
+ |4: // RA is a number, RD is not a number.
+ | bne CARG4, TISNUM, ->vmeta_comp
+ | // RA is a number, RD is an integer. Convert RD to a number.
+ |.if FPU
+ |. lwc1 FRD, LO(RD)
+ | ldc1 FRA, 0(RA)
+ | b <3
+ |. cvt.d.w FRD, FRD
+ |.else
+ |.if "ARGRD" == "CARG1"
+ |. sextw CARG1, CARG1
+ | bal ->vm_sfi2d_1
+ |. nop
+ |.else
+ |. sextw CARG2, CARG2
+ | bal ->vm_sfi2d_2
+ |. nop
+ |.endif
+ | b <3
+ |. nop
+ |.endif
+ |
+ |5: // RA is an integer, RD is not an integer
+ | sltiu AT, CARG4, LJ_TISNUM
+ | beqz AT, ->vmeta_comp
+ |. lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | // RA is an integer, RD is a number. Convert RA to a number.
+ |.if FPU
+ | lwc1 FRA, LO(RA)
+ | ldc1 FRD, 0(RD)
+ | b <3
+ | cvt.d.w FRA, FRA
+ |.else
+ |.if "ARGRA" == "CARG1"
+ | bal ->vm_sfi2d_1
+ |. sextw CARG1, CARG1
+ |.else
+ | bal ->vm_sfi2d_2
+ |. sextw CARG2, CARG2
+ |.endif
+ | b <3
+ |. nop
+ |.endif
+ |.endmacro
+ |
+ if (op == BC_ISLT) {
+ | bc_comp f20, f22, CARG1, CARG2, movz, movf, c.olt.d, ->vm_sfcmpolt
+ } else if (op == BC_ISGE) {
+ | bc_comp f20, f22, CARG1, CARG2, movn, movt, c.olt.d, ->vm_sfcmpolt
+ } else if (op == BC_ISLE) {
+ | bc_comp f22, f20, CARG2, CARG1, movn, movt, c.ult.d, ->vm_sfcmpult
+ } else {
+ | bc_comp f22, f20, CARG2, CARG1, movz, movf, c.ult.d, ->vm_sfcmpult
+ }
+ break;
+
+ case BC_ISEQV: case BC_ISNEV:
+ vk = op == BC_ISEQV;
+ | // RA = src1*8, RD = src2*8, JMP with RD = target
+ | daddu RA, BASE, RA
+ | daddiu PC, PC, 4
+ | daddu RD, BASE, RD
+ | ld CARG1, 0(RA)
+ | lhu TMP2, -4+OFS_RD(PC)
+ | ld CARG2, 0(RD)
+ | gettp CARG3, CARG1
+ | gettp CARG4, CARG2
+ | sltu AT, TISNUM, CARG3
+ | sltu TMP1, TISNUM, CARG4
+ | or AT, AT, TMP1
+ if (vk) {
+ | beqz AT, ->BC_ISEQN_Z
+ } else {
+ | beqz AT, ->BC_ISNEN_Z
+ }
+ | // Either or both types are not numbers.
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ |.if FFI
+ |. li AT, LJ_TCDATA
+ | beq CARG3, AT, ->vmeta_equal_cd
+ |.endif
+ | decode_RD4b TMP2
+ |.if FFI
+ | beq CARG4, AT, ->vmeta_equal_cd
+ |. nop
+ |.endif
+ | bne CARG1, CARG2, >2
+ |. addu TMP2, TMP2, TMP3
+ | // Tag and value are equal.
+ if (vk) {
+ |->BC_ISEQV_Z:
+ | daddu PC, PC, TMP2
+ }
+ |1:
+ | ins_next
+ |
+ |2: // Check if the tags are the same and it's a table or userdata.
+ | xor AT, CARG3, CARG4 // Same type?
+ | sltiu TMP0, CARG3, LJ_TISTABUD+1 // Table or userdata?
+ | movn TMP0, r0, AT
+ if (vk) {
+ | beqz TMP0, <1
+ } else {
+ | beqz TMP0, ->BC_ISEQV_Z // Reuse code from opposite instruction.
+ }
+ | // Different tables or userdatas. Need to check __eq metamethod.
+ | // Field metatable must be at same offset for GCtab and GCudata!
+ |. cleartp TAB:TMP1, CARG1
+ | ld TAB:TMP3, TAB:TMP1->metatable
+ if (vk) {
+ | beqz TAB:TMP3, <1 // No metatable?
+ |. nop
+ | lbu TMP3, TAB:TMP3->nomm
+ | andi TMP3, TMP3, 1<<MM_eq
+ | bnez TMP3, >1 // Or 'no __eq' flag set?
+ } else {
+ | beqz TAB:TMP3,->BC_ISEQV_Z // No metatable?
+ |. nop
+ | lbu TMP3, TAB:TMP3->nomm
+ | andi TMP3, TMP3, 1<<MM_eq
+ | bnez TMP3, ->BC_ISEQV_Z // Or 'no __eq' flag set?
+ }
+ |. nop
+ | b ->vmeta_equal // Handle __eq metamethod.
+ |. li TMP0, 1-vk // ne = 0 or 1.
+ break;
+
+ case BC_ISEQS: case BC_ISNES:
+ vk = op == BC_ISEQS;
+ | // RA = src*8, RD = str_const*8 (~), JMP with RD = target
+ | daddu RA, BASE, RA
+ | daddiu PC, PC, 4
+ | ld CARG1, 0(RA)
+ | dsubu RD, KBASE, RD
+ | lhu TMP2, -4+OFS_RD(PC)
+ | ld CARG2, -8(RD) // KBASE-8-str_const*8
+ |.if FFI
+ | gettp TMP0, CARG1
+ | li AT, LJ_TCDATA
+ |.endif
+ | li TMP1, LJ_TSTR
+ | decode_RD4b TMP2
+ |.if FFI
+ | beq TMP0, AT, ->vmeta_equal_cd
+ |.endif
+ |. settp CARG2, TMP1
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | xor TMP1, CARG1, CARG2
+ | addu TMP2, TMP2, TMP3
+ if (vk) {
+ | movn TMP2, r0, TMP1
+ } else {
+ | movz TMP2, r0, TMP1
+ }
+ | daddu PC, PC, TMP2
+ | ins_next
+ break;
+
+ case BC_ISEQN: case BC_ISNEN:
+ vk = op == BC_ISEQN;
+ | // RA = src*8, RD = num_const*8, JMP with RD = target
+ | daddu RA, BASE, RA
+ | daddu RD, KBASE, RD
+ | ld CARG1, 0(RA)
+ | ld CARG2, 0(RD)
+ | lhu TMP2, OFS_RD(PC)
+ | gettp CARG3, CARG1
+ | gettp CARG4, CARG2
+ | daddiu PC, PC, 4
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ if (vk) {
+ |->BC_ISEQN_Z:
+ } else {
+ |->BC_ISNEN_Z:
+ }
+ | bne CARG3, TISNUM, >3
+ |. decode_RD4b TMP2
+ | bne CARG4, TISNUM, >6
+ |. addu TMP2, TMP2, TMP3
+ | xor AT, CARG1, CARG2
+ if (vk) {
+ | movn TMP2, r0, AT
+ |1:
+ | daddu PC, PC, TMP2
+ |2:
+ } else {
+ | movz TMP2, r0, AT
+ |1:
+ |2:
+ | daddu PC, PC, TMP2
+ }
+ | ins_next
+ |
+ |3: // RA is not an integer.
+ | sltu AT, CARG3, TISNUM
+ |.if FFI
+ | beqz AT, >8
+ |.else
+ | beqz AT, <2
+ |.endif
+ |. addu TMP2, TMP2, TMP3
+ | sltu AT, CARG4, TISNUM
+ |.if FPU
+ | ldc1 f20, 0(RA)
+ | ldc1 f22, 0(RD)
+ |.endif
+ | beqz AT, >5
+ |. nop
+ |4: // RA and RD are both numbers.
+ |.if FPU
+ | c.eq.d f20, f22
+ | b <1
+ if (vk) {
+ |. movf TMP2, r0
+ } else {
+ |. movt TMP2, r0
+ }
+ |.else
+ | bal ->vm_sfcmpeq
+ |. nop
+ | b <1
+ if (vk) {
+ |. movz TMP2, r0, CRET1
+ } else {
+ |. movn TMP2, r0, CRET1
+ }
+ |.endif
+ |
+ |5: // RA is a number, RD is not a number.
+ |.if FFI
+ | bne CARG4, TISNUM, >9
+ |.else
+ | bne CARG4, TISNUM, <2
+ |.endif
+ | // RA is a number, RD is an integer. Convert RD to a number.
+ |.if FPU
+ |. lwc1 f22, LO(RD)
+ | b <4
+ |. cvt.d.w f22, f22
+ |.else
+ |. sextw CARG2, CARG2
+ | bal ->vm_sfi2d_2
+ |. nop
+ | b <4
+ |. nop
+ |.endif
+ |
+ |6: // RA is an integer, RD is not an integer
+ | sltu AT, CARG4, TISNUM
+ |.if FFI
+ | beqz AT, >9
+ |.else
+ | beqz AT, <2
+ |.endif
+ | // RA is an integer, RD is a number. Convert RA to a number.
+ |.if FPU
+ |. lwc1 f20, LO(RA)
+ | ldc1 f22, 0(RD)
+ | b <4
+ | cvt.d.w f20, f20
+ |.else
+ |. sextw CARG1, CARG1
+ | bal ->vm_sfi2d_1
+ |. nop
+ | b <4
+ |. nop
+ |.endif
+ |
+ |.if FFI
+ |8:
+ | li AT, LJ_TCDATA
+ | bne CARG3, AT, <2
+ |. nop
+ | b ->vmeta_equal_cd
+ |. nop
+ |9:
+ | li AT, LJ_TCDATA
+ | bne CARG4, AT, <2
+ |. nop
+ | b ->vmeta_equal_cd
+ |. nop
+ |.endif
+ break;
+
+ case BC_ISEQP: case BC_ISNEP:
+ vk = op == BC_ISEQP;
+ | // RA = src*8, RD = primitive_type*8 (~), JMP with RD = target
+ | daddu RA, BASE, RA
+ | srl TMP1, RD, 3
+ | ld TMP0, 0(RA)
+ | lhu TMP2, OFS_RD(PC)
+ | not TMP1, TMP1
+ | gettp TMP0, TMP0
+ | daddiu PC, PC, 4
+ |.if FFI
+ | li AT, LJ_TCDATA
+ | beq TMP0, AT, ->vmeta_equal_cd
+ |.endif
+ |. xor TMP0, TMP0, TMP1
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ if (vk) {
+ | movn TMP2, r0, TMP0
+ } else {
+ | movz TMP2, r0, TMP0
+ }
+ | daddu PC, PC, TMP2
+ | ins_next
+ break;
+
+ /* -- Unary test and copy ops ------------------------------------------- */
+
+ case BC_ISTC: case BC_ISFC: case BC_IST: case BC_ISF:
+ | // RA = dst*8 or unused, RD = src*8, JMP with RD = target
+ | daddu RD, BASE, RD
+ | lhu TMP2, OFS_RD(PC)
+ | ld TMP0, 0(RD)
+ | daddiu PC, PC, 4
+ | gettp TMP0, TMP0
+ | sltiu TMP0, TMP0, LJ_TISTRUECOND
+ if (op == BC_IST || op == BC_ISF) {
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ if (op == BC_IST) {
+ | movz TMP2, r0, TMP0
+ } else {
+ | movn TMP2, r0, TMP0
+ }
+ | daddu PC, PC, TMP2
+ } else {
+ | ld CRET1, 0(RD)
+ if (op == BC_ISTC) {
+ | beqz TMP0, >1
+ } else {
+ | bnez TMP0, >1
+ }
+ |. daddu RA, BASE, RA
+ | decode_RD4b TMP2
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | addu TMP2, TMP2, TMP3
+ | sd CRET1, 0(RA)
+ | daddu PC, PC, TMP2
+ |1:
+ }
+ | ins_next
+ break;
+
+ case BC_ISTYPE:
+ | // RA = src*8, RD = -type*8
+ | daddu TMP2, BASE, RA
+ | srl TMP1, RD, 3
+ | ld TMP0, 0(TMP2)
+ | ins_next1
+ | gettp TMP0, TMP0
+ | daddu AT, TMP0, TMP1
+ | bnez AT, ->vmeta_istype
+ |. ins_next2
+ break;
+ case BC_ISNUM:
+ | // RA = src*8, RD = -(TISNUM-1)*8
+ | daddu TMP2, BASE, RA
+ | ld TMP0, 0(TMP2)
+ | ins_next1
+ | checknum TMP0, ->vmeta_istype
+ |. ins_next2
+ break;
+
+ /* -- Unary ops --------------------------------------------------------- */
+
+ case BC_MOV:
+ | // RA = dst*8, RD = src*8
+ | daddu RD, BASE, RD
+ | daddu RA, BASE, RA
+ | ld CRET1, 0(RD)
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ break;
+ case BC_NOT:
+ | // RA = dst*8, RD = src*8
+ | daddu RD, BASE, RD
+ | daddu RA, BASE, RA
+ | ld TMP0, 0(RD)
+ | li AT, LJ_TTRUE
+ | gettp TMP0, TMP0
+ | sltu TMP0, AT, TMP0
+ | addiu TMP0, TMP0, 1
+ | dsll TMP0, TMP0, 47
+ | not TMP0, TMP0
+ | ins_next1
+ | sd TMP0, 0(RA)
+ | ins_next2
+ break;
+ case BC_UNM:
+ | // RA = dst*8, RD = src*8
+ | daddu RB, BASE, RD
+ | ld CARG1, 0(RB)
+ | daddu RA, BASE, RA
+ | gettp CARG3, CARG1
+ | bne CARG3, TISNUM, >2
+ |. lui TMP1, 0x8000
+ | sextw CARG1, CARG1
+ | beq CARG1, TMP1, ->vmeta_unm // Meta handler deals with -2^31.
+ |. negu CARG1, CARG1
+ | zextw CARG1, CARG1
+ | settp CARG1, TISNUM
+ |1:
+ | ins_next1
+ | sd CARG1, 0(RA)
+ | ins_next2
+ |2:
+ | sltiu AT, CARG3, LJ_TISNUM
+ | beqz AT, ->vmeta_unm
+ |. dsll TMP1, TMP1, 32
+ | b <1
+ |. xor CARG1, CARG1, TMP1
+ break;
+ case BC_LEN:
+ | // RA = dst*8, RD = src*8
+ | daddu CARG2, BASE, RD
+ | daddu RA, BASE, RA
+ | ld TMP0, 0(CARG2)
+ | gettp TMP1, TMP0
+ | daddiu AT, TMP1, -LJ_TSTR
+ | bnez AT, >2
+ |. cleartp STR:CARG1, TMP0
+ | lw CRET1, STR:CARG1->len
+ |1:
+ | settp CRET1, TISNUM
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ |2:
+ | daddiu AT, TMP1, -LJ_TTAB
+ | bnez AT, ->vmeta_len
+ |. nop
+#if LJ_52
+ | ld TAB:TMP2, TAB:CARG1->metatable
+ | bnez TAB:TMP2, >9
+ |. nop
+ |3:
+#endif
+ |->BC_LEN_Z:
+ | load_got lj_tab_len
+ | call_intern lj_tab_len // (GCtab *t)
+ |. nop
+ | // Returns uint32_t (but less than 2^31).
+ | b <1
+ |. nop
+#if LJ_52
+ |9:
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_len
+ | bnez TMP0, <3 // 'no __len' flag set: done.
+ |. nop
+ | b ->vmeta_len
+ |. nop
+#endif
+ break;
+
+ /* -- Binary ops -------------------------------------------------------- */
+
+ |.macro fpmod, a, b, c
+ | bal ->vm_floor // floor(b/c)
+ |. div.d FARG1, b, c
+ | mul.d a, FRET1, c
+ | sub.d a, b, a // b - floor(b/c)*c
+ |.endmacro
+
+ |.macro sfpmod
+ | daddiu sp, sp, -16
+ |
+ | load_got __divdf3
+ | sd CARG1, 0(sp)
+ | call_extern
+ |. sd CARG2, 8(sp)
+ |
+ | load_got floor
+ | call_extern
+ |. move CARG1, CRET1
+ |
+ | load_got __muldf3
+ | move CARG1, CRET1
+ | call_extern
+ |. ld CARG2, 8(sp)
+ |
+ | load_got __subdf3
+ | ld CARG1, 0(sp)
+ | call_extern
+ |. move CARG2, CRET1
+ |
+ | daddiu sp, sp, 16
+ |.endmacro
+
+ |.macro ins_arithpre, label
+ ||vk = ((int)op - BC_ADDVN) / (BC_ADDNV-BC_ADDVN);
+ | // RA = dst*8, RB = src1*8, RC = src2*8 | num_const*8
+ ||switch (vk) {
+ ||case 0:
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | // RA = dst*8, RB = src1*8, RC = num_const*8
+ | daddu RB, BASE, RB
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. daddu RC, KBASE, RC
+ || break;
+ ||case 1:
+ | decode_RB8a RC, INS
+ | decode_RB8b RC
+ | decode_RDtoRC8 RB, RD
+ | // RA = dst*8, RB = num_const*8, RC = src1*8
+ | daddu RC, BASE, RC
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. daddu RB, KBASE, RB
+ || break;
+ ||default:
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | // RA = dst*8, RB = src1*8, RC = src2*8
+ | daddu RB, BASE, RB
+ |.if "label" ~= "none"
+ | b label
+ |.endif
+ |. daddu RC, BASE, RC
+ || break;
+ ||}
+ |.endmacro
+ |
+ |.macro ins_arith, intins, fpins, fpcall, label
+ | ins_arithpre none
+ |
+ |.if "label" ~= "none"
+ |label:
+ |.endif
+ |
+ |// Used in 5.
+ | ld CARG1, 0(RB)
+ | ld CARG2, 0(RC)
+ | gettp TMP0, CARG1
+ | gettp TMP1, CARG2
+ |
+ |.if "intins" ~= "div"
+ |
+ | // Check for two integers.
+ | sextw CARG3, CARG1
+ | bne TMP0, TISNUM, >5
+ |. sextw CARG4, CARG2
+ | bne TMP1, TISNUM, >5
+ |
+ |.if "intins" == "addu"
+ |. intins CRET1, CARG3, CARG4
+ | xor TMP1, CRET1, CARG3 // ((y^a) & (y^b)) < 0: overflow.
+ | xor TMP2, CRET1, CARG4
+ | and TMP1, TMP1, TMP2
+ | bltz TMP1, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ |.elif "intins" == "subu"
+ |. intins CRET1, CARG3, CARG4
+ | xor TMP1, CRET1, CARG3 // ((y^a) & (a^b)) < 0: overflow.
+ | xor TMP2, CARG3, CARG4
+ | and TMP1, TMP1, TMP2
+ | bltz TMP1, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ |.elif "intins" == "mult"
+ |. intins CARG3, CARG4
+ | mflo CRET1
+ | mfhi TMP2
+ | sra TMP1, CRET1, 31
+ | bne TMP1, TMP2, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ |.else
+ |. load_got lj_vm_modi
+ | beqz CARG4, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ | move CARG1, CARG3
+ | call_extern
+ |. move CARG2, CARG4
+ |.endif
+ |
+ | zextw CRET1, CRET1
+ | settp CRET1, TISNUM
+ | ins_next1
+ | sd CRET1, 0(RA)
+ |3:
+ | ins_next2
+ |
+ |.endif
+ |
+ |5: // Check for two numbers.
+ | .FPU ldc1 f20, 0(RB)
+ | sltu AT, TMP0, TISNUM
+ | sltu TMP0, TMP1, TISNUM
+ | .FPU ldc1 f22, 0(RC)
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ |
+ |.if FPU
+ | fpins FRET1, f20, f22
+ |.elif "fpcall" == "sfpmod"
+ | sfpmod
+ |.else
+ | load_got fpcall
+ | call_extern
+ |. nop
+ |.endif
+ |
+ | ins_next1
+ |.if "intins" ~= "div"
+ | b <3
+ |.endif
+ |.if FPU
+ |. sdc1 FRET1, 0(RA)
+ |.else
+ |. sd CRET1, 0(RA)
+ |.endif
+ |.if "intins" == "div"
+ | ins_next2
+ |.endif
+ |
+ |.endmacro
+
+ case BC_ADDVN: case BC_ADDNV: case BC_ADDVV:
+ | ins_arith addu, add.d, __adddf3, none
+ break;
+ case BC_SUBVN: case BC_SUBNV: case BC_SUBVV:
+ | ins_arith subu, sub.d, __subdf3, none
+ break;
+ case BC_MULVN: case BC_MULNV: case BC_MULVV:
+ | ins_arith mult, mul.d, __muldf3, none
+ break;
+ case BC_DIVVN:
+ | ins_arith div, div.d, __divdf3, ->BC_DIVVN_Z
+ break;
+ case BC_DIVNV: case BC_DIVVV:
+ | ins_arithpre ->BC_DIVVN_Z
+ break;
+ case BC_MODVN:
+ | ins_arith modi, fpmod, sfpmod, ->BC_MODVN_Z
+ break;
+ case BC_MODNV: case BC_MODVV:
+ | ins_arithpre ->BC_MODVN_Z
+ break;
+ case BC_POW:
+ | ins_arithpre none
+ | ld CARG1, 0(RB)
+ | ld CARG2, 0(RC)
+ | gettp TMP0, CARG1
+ | gettp TMP1, CARG2
+ | sltiu TMP0, TMP0, LJ_TISNUM
+ | sltiu TMP1, TMP1, LJ_TISNUM
+ | and AT, TMP0, TMP1
+ | load_got pow
+ | beqz AT, ->vmeta_arith
+ |. daddu RA, BASE, RA
+ |.if FPU
+ | ldc1 FARG1, 0(RB)
+ | ldc1 FARG2, 0(RC)
+ |.endif
+ | call_extern
+ |. nop
+ | ins_next1
+ |.if FPU
+ | sdc1 FRET1, 0(RA)
+ |.else
+ | sd CRET1, 0(RA)
+ |.endif
+ | ins_next2
+ break;
+
+ case BC_CAT:
+ | // RA = dst*8, RB = src_start*8, RC = src_end*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | dsubu CARG3, RC, RB
+ | sd BASE, L->base
+ | daddu CARG2, BASE, RC
+ | move MULTRES, RB
+ |->BC_CAT_Z:
+ | load_got lj_meta_cat
+ | srl CARG3, CARG3, 3
+ | sd PC, SAVE_PC
+ | call_intern lj_meta_cat // (lua_State *L, TValue *top, int left)
+ |. move CARG1, L
+ | // Returns NULL (finished) or TValue * (metamethod).
+ | bnez CRET1, ->vmeta_binop
+ |. ld BASE, L->base
+ | daddu RB, BASE, MULTRES
+ | ld CRET1, 0(RB)
+ | daddu RA, BASE, RA
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ break;
+
+ /* -- Constant ops ------------------------------------------------------ */
+
+ case BC_KSTR:
+ | // RA = dst*8, RD = str_const*8 (~)
+ | dsubu TMP1, KBASE, RD
+ | ins_next1
+ | li TMP2, LJ_TSTR
+ | ld TMP0, -8(TMP1) // KBASE-8-str_const*8
+ | daddu RA, BASE, RA
+ | settp TMP0, TMP2
+ | sd TMP0, 0(RA)
+ | ins_next2
+ break;
+ case BC_KCDATA:
+ |.if FFI
+ | // RA = dst*8, RD = cdata_const*8 (~)
+ | dsubu TMP1, KBASE, RD
+ | ins_next1
+ | ld TMP0, -8(TMP1) // KBASE-8-cdata_const*8
+ | li TMP2, LJ_TCDATA
+ | daddu RA, BASE, RA
+ | settp TMP0, TMP2
+ | sd TMP0, 0(RA)
+ | ins_next2
+ |.endif
+ break;
+ case BC_KSHORT:
+ | // RA = dst*8, RD = int16_literal*8
+ | sra RD, INS, 16
+ | daddu RA, BASE, RA
+ | zextw RD, RD
+ | ins_next1
+ | settp RD, TISNUM
+ | sd RD, 0(RA)
+ | ins_next2
+ break;
+ case BC_KNUM:
+ | // RA = dst*8, RD = num_const*8
+ | daddu RD, KBASE, RD
+ | daddu RA, BASE, RA
+ | ld CRET1, 0(RD)
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ break;
+ case BC_KPRI:
+ | // RA = dst*8, RD = primitive_type*8 (~)
+ | daddu RA, BASE, RA
+ | dsll TMP0, RD, 44
+ | not TMP0, TMP0
+ | ins_next1
+ | sd TMP0, 0(RA)
+ | ins_next2
+ break;
+ case BC_KNIL:
+ | // RA = base*8, RD = end*8
+ | daddu RA, BASE, RA
+ | sd TISNIL, 0(RA)
+ | daddiu RA, RA, 8
+ | daddu RD, BASE, RD
+ |1:
+ | sd TISNIL, 0(RA)
+ | slt AT, RA, RD
+ | bnez AT, <1
+ |. daddiu RA, RA, 8
+ | ins_next_
+ break;
+
+ /* -- Upvalue and function ops ------------------------------------------ */
+
+ case BC_UGET:
+ | // RA = dst*8, RD = uvnum*8
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | daddu RA, BASE, RA
+ | cleartp LFUNC:RB
+ | daddu RD, RD, LFUNC:RB
+ | ld UPVAL:RB, LFUNC:RD->uvptr
+ | ins_next1
+ | ld TMP1, UPVAL:RB->v
+ | ld CRET1, 0(TMP1)
+ | sd CRET1, 0(RA)
+ | ins_next2
+ break;
+ case BC_USETV:
+ | // RA = uvnum*8, RD = src*8
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | daddu RD, BASE, RD
+ | cleartp LFUNC:RB
+ | daddu RA, RA, LFUNC:RB
+ | ld UPVAL:RB, LFUNC:RA->uvptr
+ | ld CRET1, 0(RD)
+ | lbu TMP3, UPVAL:RB->marked
+ | ld CARG2, UPVAL:RB->v
+ | andi TMP3, TMP3, LJ_GC_BLACK // isblack(uv)
+ | lbu TMP0, UPVAL:RB->closed
+ | gettp TMP2, CRET1
+ | sd CRET1, 0(CARG2)
+ | li AT, LJ_GC_BLACK|1
+ | or TMP3, TMP3, TMP0
+ | beq TMP3, AT, >2 // Upvalue is closed and black?
+ |. daddiu TMP2, TMP2, -(LJ_TNUMX+1)
+ |1:
+ | ins_next
+ |
+ |2: // Check if new value is collectable.
+ | sltiu AT, TMP2, LJ_TISGCV - (LJ_TNUMX+1)
+ | beqz AT, <1 // tvisgcv(v)
+ |. cleartp GCOBJ:CRET1, CRET1
+ | lbu TMP3, GCOBJ:CRET1->gch.marked
+ | andi TMP3, TMP3, LJ_GC_WHITES // iswhite(v)
+ | beqz TMP3, <1
+ |. load_got lj_gc_barrieruv
+ | // Crossed a write barrier. Move the barrier forward.
+ | call_intern lj_gc_barrieruv // (global_State *g, TValue *tv)
+ |. daddiu CARG1, DISPATCH, GG_DISP2G
+ | b <1
+ |. nop
+ break;
+ case BC_USETS:
+ | // RA = uvnum*8, RD = str_const*8 (~)
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | dsubu TMP1, KBASE, RD
+ | cleartp LFUNC:RB
+ | daddu RA, RA, LFUNC:RB
+ | ld UPVAL:RB, LFUNC:RA->uvptr
+ | ld STR:TMP1, -8(TMP1) // KBASE-8-str_const*8
+ | lbu TMP2, UPVAL:RB->marked
+ | ld CARG2, UPVAL:RB->v
+ | lbu TMP3, STR:TMP1->marked
+ | andi AT, TMP2, LJ_GC_BLACK // isblack(uv)
+ | lbu TMP2, UPVAL:RB->closed
+ | li TMP0, LJ_TSTR
+ | settp TMP1, TMP0
+ | bnez AT, >2
+ |. sd TMP1, 0(CARG2)
+ |1:
+ | ins_next
+ |
+ |2: // Check if string is white and ensure upvalue is closed.
+ | beqz TMP2, <1
+ |. andi AT, TMP3, LJ_GC_WHITES // iswhite(str)
+ | beqz AT, <1
+ |. load_got lj_gc_barrieruv
+ | // Crossed a write barrier. Move the barrier forward.
+ | call_intern lj_gc_barrieruv // (global_State *g, TValue *tv)
+ |. daddiu CARG1, DISPATCH, GG_DISP2G
+ | b <1
+ |. nop
+ break;
+ case BC_USETN:
+ | // RA = uvnum*8, RD = num_const*8
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | daddu RD, KBASE, RD
+ | cleartp LFUNC:RB
+ | daddu RA, RA, LFUNC:RB
+ | ld UPVAL:RB, LFUNC:RA->uvptr
+ | ld CRET1, 0(RD)
+ | ld TMP1, UPVAL:RB->v
+ | ins_next1
+ | sd CRET1, 0(TMP1)
+ | ins_next2
+ break;
+ case BC_USETP:
+ | // RA = uvnum*8, RD = primitive_type*8 (~)
+ | ld LFUNC:RB, FRAME_FUNC(BASE)
+ | dsll TMP0, RD, 44
+ | cleartp LFUNC:RB
+ | daddu RA, RA, LFUNC:RB
+ | not TMP0, TMP0
+ | ld UPVAL:RB, LFUNC:RA->uvptr
+ | ins_next1
+ | ld TMP1, UPVAL:RB->v
+ | sd TMP0, 0(TMP1)
+ | ins_next2
+ break;
+
+ case BC_UCLO:
+ | // RA = level*8, RD = target
+ | ld TMP2, L->openupval
+ | branch_RD // Do this first since RD is not saved.
+ | load_got lj_func_closeuv
+ | sd BASE, L->base
+ | beqz TMP2, >1
+ |. move CARG1, L
+ | call_intern lj_func_closeuv // (lua_State *L, TValue *level)
+ |. daddu CARG2, BASE, RA
+ | ld BASE, L->base
+ |1:
+ | ins_next
+ break;
+
+ case BC_FNEW:
+ | // RA = dst*8, RD = proto_const*8 (~) (holding function prototype)
+ | load_got lj_func_newL_gc
+ | dsubu TMP1, KBASE, RD
+ | ld CARG3, FRAME_FUNC(BASE)
+ | ld CARG2, -8(TMP1) // KBASE-8-tab_const*8
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | cleartp CARG3
+ | // (lua_State *L, GCproto *pt, GCfuncL *parent)
+ | call_intern lj_func_newL_gc
+ |. move CARG1, L
+ | // Returns GCfuncL *.
+ | li TMP0, LJ_TFUNC
+ | ld BASE, L->base
+ | ins_next1
+ | settp CRET1, TMP0
+ | daddu RA, BASE, RA
+ | sd CRET1, 0(RA)
+ | ins_next2
+ break;
+
+ /* -- Table ops --------------------------------------------------------- */
+
+ case BC_TNEW:
+ case BC_TDUP:
+ | // RA = dst*8, RD = (hbits|asize)*8 | tab_const*8 (~)
+ | ld TMP0, DISPATCH_GL(gc.total)(DISPATCH)
+ | ld TMP1, DISPATCH_GL(gc.threshold)(DISPATCH)
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | sltu AT, TMP0, TMP1
+ | beqz AT, >5
+ |1:
+ if (op == BC_TNEW) {
+ | load_got lj_tab_new
+ | srl CARG2, RD, 3
+ | andi CARG2, CARG2, 0x7ff
+ | li TMP0, 0x801
+ | addiu AT, CARG2, -0x7ff
+ | srl CARG3, RD, 14
+ | movz CARG2, TMP0, AT
+ | // (lua_State *L, int32_t asize, uint32_t hbits)
+ | call_intern lj_tab_new
+ |. move CARG1, L
+ | // Returns Table *.
+ } else {
+ | load_got lj_tab_dup
+ | dsubu TMP1, KBASE, RD
+ | move CARG1, L
+ | call_intern lj_tab_dup // (lua_State *L, Table *kt)
+ |. ld CARG2, -8(TMP1) // KBASE-8-str_const*8
+ | // Returns Table *.
+ }
+ | li TMP0, LJ_TTAB
+ | ld BASE, L->base
+ | ins_next1
+ | daddu RA, BASE, RA
+ | settp CRET1, TMP0
+ | sd CRET1, 0(RA)
+ | ins_next2
+ |5:
+ | load_got lj_gc_step_fixtop
+ | move MULTRES, RD
+ | call_intern lj_gc_step_fixtop // (lua_State *L)
+ |. move CARG1, L
+ | b <1
+ |. move RD, MULTRES
+ break;
+
+ case BC_GGET:
+ | // RA = dst*8, RD = str_const*8 (~)
+ case BC_GSET:
+ | // RA = src*8, RD = str_const*8 (~)
+ | ld LFUNC:TMP2, FRAME_FUNC(BASE)
+ | dsubu TMP1, KBASE, RD
+ | ld STR:RC, -8(TMP1) // KBASE-8-str_const*8
+ | cleartp LFUNC:TMP2
+ | ld TAB:RB, LFUNC:TMP2->env
+ if (op == BC_GGET) {
+ | b ->BC_TGETS_Z
+ } else {
+ | b ->BC_TSETS_Z
+ }
+ |. daddu RA, BASE, RA
+ break;
+
+ case BC_TGETV:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | daddu CARG2, BASE, RB
+ | daddu CARG3, BASE, RC
+ | ld TAB:RB, 0(CARG2)
+ | ld TMP2, 0(CARG3)
+ | daddu RA, BASE, RA
+ | checktab TAB:RB, ->vmeta_tgetv
+ | gettp TMP3, TMP2
+ | bne TMP3, TISNUM, >5 // Integer key?
+ |. lw TMP0, TAB:RB->asize
+ | sextw TMP2, TMP2
+ | ld TMP1, TAB:RB->array
+ | sltu AT, TMP2, TMP0
+ | sll TMP2, TMP2, 3
+ | beqz AT, ->vmeta_tgetv // Integer key and in array part?
+ |. daddu TMP2, TMP1, TMP2
+ | ld AT, 0(TMP2)
+ | beq AT, TISNIL, >2
+ |. ld CRET1, 0(TMP2)
+ |1:
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ |
+ |2: // Check for __index if table value is nil.
+ | ld TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_index
+ | bnez TMP0, <1 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgetv
+ |. nop
+ |
+ |5:
+ | li AT, LJ_TSTR
+ | bne TMP3, AT, ->vmeta_tgetv
+ |. cleartp RC, TMP2
+ | b ->BC_TGETS_Z // String key?
+ |. nop
+ break;
+ case BC_TGETS:
+ | // RA = dst*8, RB = table*8, RC = str_const*8 (~)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RC8a RC, INS
+ | daddu CARG2, BASE, RB
+ | decode_RC8b RC
+ | ld TAB:RB, 0(CARG2)
+ | dsubu CARG3, KBASE, RC
+ | daddu RA, BASE, RA
+ | ld STR:RC, -8(CARG3) // KBASE-8-str_const*8
+ | checktab TAB:RB, ->vmeta_tgets1
+ |->BC_TGETS_Z:
+ | // TAB:RB = GCtab *, STR:RC = GCstr *, RA = dst*8
+ | lw TMP0, TAB:RB->hmask
+ | lw TMP1, STR:RC->hash
+ | ld NODE:TMP2, TAB:RB->node
+ | and TMP1, TMP1, TMP0 // idx = str->hash & tab->hmask
+ | sll TMP0, TMP1, 5
+ | sll TMP1, TMP1, 3
+ | subu TMP1, TMP0, TMP1
+ | li TMP3, LJ_TSTR
+ | daddu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ | settp STR:RC, TMP3 // Tagged key to look for.
+ |1:
+ | ld CARG1, NODE:TMP2->key
+ | ld CRET1, NODE:TMP2->val
+ | ld NODE:TMP1, NODE:TMP2->next
+ | bne CARG1, RC, >4
+ |. ld TAB:TMP3, TAB:RB->metatable
+ | beq CRET1, TISNIL, >5 // Key found, but nil value?
+ |. nop
+ |3:
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ |
+ |4: // Follow hash chain.
+ | bnez NODE:TMP1, <1
+ |. move NODE:TMP2, NODE:TMP1
+ | // End of hash chain: key not found, nil result.
+ |
+ |5: // Check for __index if table value is nil.
+ | beqz TAB:TMP3, <3 // No metatable: done.
+ |. move CRET1, TISNIL
+ | lbu TMP0, TAB:TMP3->nomm
+ | andi TMP0, TMP0, 1<<MM_index
+ | bnez TMP0, <3 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgets
+ |. nop
+ break;
+ case BC_TGETB:
+ | // RA = dst*8, RB = table*8, RC = index*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | daddu CARG2, BASE, RB
+ | decode_RDtoRC8 RC, RD
+ | ld TAB:RB, 0(CARG2)
+ | daddu RA, BASE, RA
+ | srl TMP0, RC, 3
+ | checktab TAB:RB, ->vmeta_tgetb
+ | lw TMP1, TAB:RB->asize
+ | ld TMP2, TAB:RB->array
+ | sltu AT, TMP0, TMP1
+ | beqz AT, ->vmeta_tgetb
+ |. daddu RC, TMP2, RC
+ | ld AT, 0(RC)
+ | beq AT, TISNIL, >5
+ |. ld CRET1, 0(RC)
+ |1:
+ | ins_next1
+ | sd CRET1, 0(RA)
+ | ins_next2
+ |
+ |5: // Check for __index if table value is nil.
+ | ld TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP1, TAB:TMP2->nomm
+ | andi TMP1, TMP1, 1<<MM_index
+ | bnez TMP1, <1 // 'no __index' flag set: done.
+ |. nop
+ | b ->vmeta_tgetb // Caveat: preserve TMP0 and CARG2!
+ |. nop
+ break;
+ case BC_TGETR:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | daddu RB, BASE, RB
+ | daddu RC, BASE, RC
+ | ld TAB:CARG1, 0(RB)
+ | lw CARG2, LO(RC)
+ | daddu RA, BASE, RA
+ | cleartp TAB:CARG1
+ | lw TMP0, TAB:CARG1->asize
+ | ld TMP1, TAB:CARG1->array
+ | sltu AT, CARG2, TMP0
+ | sll TMP2, CARG2, 3
+ | beqz AT, ->vmeta_tgetr // In array part?
+ |. daddu CRET1, TMP1, TMP2
+ | ld CARG2, 0(CRET1)
+ |->BC_TGETR_Z:
+ | ins_next1
+ | sd CARG2, 0(RA)
+ | ins_next2
+ break;
+
+ case BC_TSETV:
+ | // RA = src*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | daddu CARG2, BASE, RB
+ | daddu CARG3, BASE, RC
+ | ld RB, 0(CARG2)
+ | ld TMP2, 0(CARG3)
+ | daddu RA, BASE, RA
+ | checktab RB, ->vmeta_tsetv
+ | checkint TMP2, >5
+ |. sextw RC, TMP2
+ | lw TMP0, TAB:RB->asize
+ | ld TMP1, TAB:RB->array
+ | sltu AT, RC, TMP0
+ | sll TMP2, RC, 3
+ | beqz AT, ->vmeta_tsetv // Integer key and in array part?
+ |. daddu TMP1, TMP1, TMP2
+ | ld TMP0, 0(TMP1)
+ | lbu TMP3, TAB:RB->marked
+ | beq TMP0, TISNIL, >3
+ |. ld CRET1, 0(RA)
+ |1:
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | bnez AT, >7
+ |. sd CRET1, 0(TMP1)
+ |2:
+ | ins_next
+ |
+ |3: // Check for __newindex if previous value is nil.
+ | ld TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP2, TAB:TMP2->nomm
+ | andi TMP2, TMP2, 1<<MM_newindex
+ | bnez TMP2, <1 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsetv
+ |. nop
+ |
+ |5:
+ | gettp AT, TMP2
+ | daddiu AT, AT, -LJ_TSTR
+ | bnez AT, ->vmeta_tsetv
+ |. nop
+ | b ->BC_TSETS_Z // String key?
+ |. cleartp STR:RC, TMP2
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <2
+ break;
+ case BC_TSETS:
+ | // RA = src*8, RB = table*8, RC = str_const*8 (~)
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | daddu CARG2, BASE, RB
+ | decode_RC8a RC, INS
+ | ld TAB:RB, 0(CARG2)
+ | decode_RC8b RC
+ | dsubu CARG3, KBASE, RC
+ | ld RC, -8(CARG3) // KBASE-8-str_const*8
+ | daddu RA, BASE, RA
+ | cleartp STR:RC
+ | checktab TAB:RB, ->vmeta_tsets1
+ |->BC_TSETS_Z:
+ | // TAB:RB = GCtab *, STR:RC = GCstr *, RA = BASE+src*8
+ | lw TMP0, TAB:RB->hmask
+ | lw TMP1, STR:RC->hash
+ | ld NODE:TMP2, TAB:RB->node
+ | sb r0, TAB:RB->nomm // Clear metamethod cache.
+ | and TMP1, TMP1, TMP0 // idx = str->hash & tab->hmask
+ | sll TMP0, TMP1, 5
+ | sll TMP1, TMP1, 3
+ | subu TMP1, TMP0, TMP1
+ | li TMP3, LJ_TSTR
+ | daddu NODE:TMP2, NODE:TMP2, TMP1 // node = tab->node + (idx*32-idx*8)
+ | settp STR:RC, TMP3 // Tagged key to look for.
+ |.if FPU
+ | ldc1 f20, 0(RA)
+ |.else
+ | ld CRET1, 0(RA)
+ |.endif
+ |1:
+ | ld TMP0, NODE:TMP2->key
+ | ld CARG2, NODE:TMP2->val
+ | ld NODE:TMP1, NODE:TMP2->next
+ | bne TMP0, RC, >5
+ |. lbu TMP3, TAB:RB->marked
+ | beq CARG2, TISNIL, >4 // Key found, but nil value?
+ |. ld TAB:TMP0, TAB:RB->metatable
+ |2:
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | bnez AT, >7
+ |.if FPU
+ |. sdc1 f20, NODE:TMP2->val
+ |.else
+ |. sd CRET1, NODE:TMP2->val
+ |.endif
+ |3:
+ | ins_next
+ |
+ |4: // Check for __newindex if previous value is nil.
+ | beqz TAB:TMP0, <2 // No metatable: done.
+ |. nop
+ | lbu TMP0, TAB:TMP0->nomm
+ | andi TMP0, TMP0, 1<<MM_newindex
+ | bnez TMP0, <2 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsets
+ |. nop
+ |
+ |5: // Follow hash chain.
+ | bnez NODE:TMP1, <1
+ |. move NODE:TMP2, NODE:TMP1
+ | // End of hash chain: key not found, add a new one
+ |
+ | // But check for __newindex first.
+ | ld TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, >6 // No metatable: continue.
+ |. daddiu CARG3, DISPATCH, DISPATCH_GL(tmptv)
+ | lbu TMP0, TAB:TMP2->nomm
+ | andi TMP0, TMP0, 1<<MM_newindex
+ | beqz TMP0, ->vmeta_tsets // 'no __newindex' flag NOT set: check.
+ |6:
+ | load_got lj_tab_newkey
+ | sd RC, 0(CARG3)
+ | sd BASE, L->base
+ | move CARG2, TAB:RB
+ | sd PC, SAVE_PC
+ | call_intern lj_tab_newkey // (lua_State *L, GCtab *t, TValue *k
+ |. move CARG1, L
+ | // Returns TValue *.
+ | ld BASE, L->base
+ |.if FPU
+ | b <3 // No 2nd write barrier needed.
+ |. sdc1 f20, 0(CRET1)
+ |.else
+ | ld CARG1, 0(RA)
+ | b <3 // No 2nd write barrier needed.
+ |. sd CARG1, 0(CRET1)
+ |.endif
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <3
+ break;
+ case BC_TSETB:
+ | // RA = src*8, RB = table*8, RC = index*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | daddu CARG2, BASE, RB
+ | decode_RDtoRC8 RC, RD
+ | ld TAB:RB, 0(CARG2)
+ | daddu RA, BASE, RA
+ | srl TMP0, RC, 3
+ | checktab RB, ->vmeta_tsetb
+ | lw TMP1, TAB:RB->asize
+ | ld TMP2, TAB:RB->array
+ | sltu AT, TMP0, TMP1
+ | beqz AT, ->vmeta_tsetb
+ |. daddu RC, TMP2, RC
+ | ld TMP1, 0(RC)
+ | lbu TMP3, TAB:RB->marked
+ | beq TMP1, TISNIL, >5
+ |1:
+ |. ld CRET1, 0(RA)
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | bnez AT, >7
+ |. sd CRET1, 0(RC)
+ |2:
+ | ins_next
+ |
+ |5: // Check for __newindex if previous value is nil.
+ | ld TAB:TMP2, TAB:RB->metatable
+ | beqz TAB:TMP2, <1 // No metatable: done.
+ |. nop
+ | lbu TMP1, TAB:TMP2->nomm
+ | andi TMP1, TMP1, 1<<MM_newindex
+ | bnez TMP1, <1 // 'no __newindex' flag set: done.
+ |. nop
+ | b ->vmeta_tsetb // Caveat: preserve TMP0 and CARG2!
+ |. nop
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:RB, TMP3, TMP0, <2
+ break;
+ case BC_TSETR:
+ | // RA = dst*8, RB = table*8, RC = key*8
+ | decode_RB8a RB, INS
+ | decode_RB8b RB
+ | decode_RDtoRC8 RC, RD
+ | daddu CARG1, BASE, RB
+ | daddu CARG3, BASE, RC
+ | ld TAB:CARG2, 0(CARG1)
+ | lw CARG3, LO(CARG3)
+ | cleartp TAB:CARG2
+ | lbu TMP3, TAB:CARG2->marked
+ | lw TMP0, TAB:CARG2->asize
+ | ld TMP1, TAB:CARG2->array
+ | andi AT, TMP3, LJ_GC_BLACK // isblack(table)
+ | bnez AT, >7
+ |. daddu RA, BASE, RA
+ |2:
+ | sltu AT, CARG3, TMP0
+ | sll TMP2, CARG3, 3
+ | beqz AT, ->vmeta_tsetr // In array part?
+ |. daddu CRET1, TMP1, TMP2
+ |->BC_TSETR_Z:
+ | ld CARG1, 0(RA)
+ | ins_next1
+ | sd CARG1, 0(CRET1)
+ | ins_next2
+ |
+ |7: // Possible table write barrier for the value. Skip valiswhite check.
+ | barrierback TAB:CARG2, TMP3, TMP0, <2
+ break;
+
+ case BC_TSETM:
+ | // RA = base*8 (table at base-1), RD = num_const*8 (start index)
+ | daddu RA, BASE, RA
+ |1:
+ | daddu TMP3, KBASE, RD
+ | ld TAB:CARG2, -8(RA) // Guaranteed to be a table.
+ | addiu TMP0, MULTRES, -8
+ | lw TMP3, LO(TMP3) // Integer constant is in lo-word.
+ | beqz TMP0, >4 // Nothing to copy?
+ |. srl CARG3, TMP0, 3
+ | cleartp CARG2
+ | addu CARG3, CARG3, TMP3
+ | lw TMP2, TAB:CARG2->asize
+ | sll TMP1, TMP3, 3
+ | lbu TMP3, TAB:CARG2->marked
+ | ld CARG1, TAB:CARG2->array
+ | sltu AT, TMP2, CARG3
+ | bnez AT, >5
+ |. daddu TMP2, RA, TMP0
+ | daddu TMP1, TMP1, CARG1
+ | andi TMP0, TMP3, LJ_GC_BLACK // isblack(table)
+ |3: // Copy result slots to table.
+ | ld CRET1, 0(RA)
+ | daddiu RA, RA, 8
+ | sltu AT, RA, TMP2
+ | sd CRET1, 0(TMP1)
+ | bnez AT, <3
+ |. daddiu TMP1, TMP1, 8
+ | bnez TMP0, >7
+ |. nop
+ |4:
+ | ins_next
+ |
+ |5: // Need to resize array part.
+ | load_got lj_tab_reasize
+ | sd BASE, L->base
+ | sd PC, SAVE_PC
+ | move BASE, RD
+ | call_intern lj_tab_reasize // (lua_State *L, GCtab *t, int nasize)
+ |. move CARG1, L
+ | // Must not reallocate the stack.
+ | move RD, BASE
+ | b <1
+ |. ld BASE, L->base // Reload BASE for lack of a saved register.
+ |
+ |7: // Possible table write barrier for any value. Skip valiswhite check.
+ | barrierback TAB:CARG2, TMP3, TMP0, <4
+ break;
+
+ /* -- Calls and vararg handling ----------------------------------------- */
+
+ case BC_CALLM:
+ | // RA = base*8, (RB = (nresults+1)*8,) RC = extra_nargs*8
+ | decode_RDtoRC8 NARGS8:RC, RD
+ | b ->BC_CALL_Z
+ |. addu NARGS8:RC, NARGS8:RC, MULTRES
+ break;
+ case BC_CALL:
+ | // RA = base*8, (RB = (nresults+1)*8,) RC = (nargs+1)*8
+ | decode_RDtoRC8 NARGS8:RC, RD
+ |->BC_CALL_Z:
+ | move TMP2, BASE
+ | daddu BASE, BASE, RA
+ | ld LFUNC:RB, 0(BASE)
+ | daddiu BASE, BASE, 16
+ | addiu NARGS8:RC, NARGS8:RC, -8
+ | checkfunc RB, ->vmeta_call
+ | ins_call
+ break;
+
+ case BC_CALLMT:
+ | // RA = base*8, (RB = 0,) RC = extra_nargs*8
+ | addu NARGS8:RD, NARGS8:RD, MULTRES // BC_CALLT gets RC from RD.
+ | // Fall through. Assumes BC_CALLT follows.
+ break;
+ case BC_CALLT:
+ | // RA = base*8, (RB = 0,) RC = (nargs+1)*8
+ | daddu RA, BASE, RA
+ | ld RB, 0(RA)
+ | move NARGS8:RC, RD
+ | ld TMP1, FRAME_PC(BASE)
+ | daddiu RA, RA, 16
+ | addiu NARGS8:RC, NARGS8:RC, -8
+ | checktp CARG3, RB, -LJ_TFUNC, ->vmeta_callt
+ |->BC_CALLT_Z:
+ | andi TMP0, TMP1, FRAME_TYPE // Caveat: preserve TMP0 until the 'or'.
+ | lbu TMP3, LFUNC:CARG3->ffid
+ | bnez TMP0, >7
+ |. xori TMP2, TMP1, FRAME_VARG
+ |1:
+ | sd RB, FRAME_FUNC(BASE) // Copy function down, but keep PC.
+ | sltiu AT, TMP3, 2 // (> FF_C) Calling a fast function?
+ | move TMP2, BASE
+ | move RB, CARG3
+ | beqz NARGS8:RC, >3
+ |. move TMP3, NARGS8:RC
+ |2:
+ | ld CRET1, 0(RA)
+ | daddiu RA, RA, 8
+ | addiu TMP3, TMP3, -8
+ | sd CRET1, 0(TMP2)
+ | bnez TMP3, <2
+ |. daddiu TMP2, TMP2, 8
+ |3:
+ | or TMP0, TMP0, AT
+ | beqz TMP0, >5
+ |. nop
+ |4:
+ | ins_callt
+ |
+ |5: // Tailcall to a fast function with a Lua frame below.
+ | lw INS, -4(TMP1)
+ | decode_RA8a RA, INS
+ | decode_RA8b RA
+ | dsubu TMP1, BASE, RA
+ | ld TMP1, -32(TMP1)
+ | cleartp LFUNC:TMP1
+ | ld TMP1, LFUNC:TMP1->pc
+ | b <4
+ |. ld KBASE, PC2PROTO(k)(TMP1) // Need to prepare KBASE.
+ |
+ |7: // Tailcall from a vararg function.
+ | andi AT, TMP2, FRAME_TYPEP
+ | bnez AT, <1 // Vararg frame below?
+ |. dsubu TMP2, BASE, TMP2 // Relocate BASE down.
+ | move BASE, TMP2
+ | ld TMP1, FRAME_PC(TMP2)
+ | b <1
+ |. andi TMP0, TMP1, FRAME_TYPE
+ break;
+
+ case BC_ITERC:
+ | // RA = base*8, (RB = (nresults+1)*8, RC = (nargs+1)*8 ((2+1)*8))
+ | move TMP2, BASE // Save old BASE fir vmeta_call.
+ | daddu BASE, BASE, RA
+ | ld RB, -24(BASE)
+ | ld CARG1, -16(BASE)
+ | ld CARG2, -8(BASE)
+ | li NARGS8:RC, 16 // Iterators get 2 arguments.
+ | sd RB, 0(BASE) // Copy callable.
+ | sd CARG1, 16(BASE) // Copy state.
+ | sd CARG2, 24(BASE) // Copy control var.
+ | daddiu BASE, BASE, 16
+ | checkfunc RB, ->vmeta_call
+ | ins_call
+ break;
+
+ case BC_ITERN:
+ | // RA = base*8, (RB = (nresults+1)*8, RC = (nargs+1)*8 (2+1)*8)
+ |.if JIT
+ | // NYI: add hotloop, record BC_ITERN.
+ |.endif
+ | daddu RA, BASE, RA
+ | ld TAB:RB, -16(RA)
+ | lw RC, -8+LO(RA) // Get index from control var.
+ | cleartp TAB:RB
+ | daddiu PC, PC, 4
+ | lw TMP0, TAB:RB->asize
+ | ld TMP1, TAB:RB->array
+ | dsll CARG3, TISNUM, 47
+ |1: // Traverse array part.
+ | sltu AT, RC, TMP0
+ | beqz AT, >5 // Index points after array part?
+ |. sll TMP3, RC, 3
+ | daddu TMP3, TMP1, TMP3
+ | ld CARG1, 0(TMP3)
+ | lhu RD, -4+OFS_RD(PC)
+ | or TMP2, RC, CARG3
+ | beq CARG1, TISNIL, <1 // Skip holes in array part.
+ |. addiu RC, RC, 1
+ | sd TMP2, 0(RA)
+ | sd CARG1, 8(RA)
+ | or TMP0, RC, CARG3
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | decode_RD4b RD
+ | daddu RD, RD, TMP3
+ | sw TMP0, -8+LO(RA) // Update control var.
+ | daddu PC, PC, RD
+ |3:
+ | ins_next
+ |
+ |5: // Traverse hash part.
+ | lw TMP1, TAB:RB->hmask
+ | subu RC, RC, TMP0
+ | ld TMP2, TAB:RB->node
+ |6:
+ | sltu AT, TMP1, RC // End of iteration? Branch to ITERL+1.
+ | bnez AT, <3
+ |. sll TMP3, RC, 5
+ | sll RB, RC, 3
+ | subu TMP3, TMP3, RB
+ | daddu NODE:TMP3, TMP3, TMP2
+ | ld CARG1, 0(NODE:TMP3)
+ | lhu RD, -4+OFS_RD(PC)
+ | beq CARG1, TISNIL, <6 // Skip holes in hash part.
+ |. addiu RC, RC, 1
+ | ld CARG2, NODE:TMP3->key
+ | lui TMP3, (-(BCBIAS_J*4 >> 16) & 65535)
+ | sd CARG1, 8(RA)
+ | addu RC, RC, TMP0
+ | decode_RD4b RD
+ | addu RD, RD, TMP3
+ | sd CARG2, 0(RA)
+ | daddu PC, PC, RD
+ | b <3
+ |. sw RC, -8+LO(RA) // Update control var.
+ break;
+
+ case BC_ISNEXT:
+ | // RA = base*8, RD = target (points to ITERN)
+ | daddu RA, BASE, RA
+ | srl TMP0, RD, 1
+ | ld CFUNC:CARG1, -24(RA)
+ | daddu TMP0, PC, TMP0
+ | ld CARG2, -16(RA)
+ | ld CARG3, -8(RA)
+ | lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | checkfunc CFUNC:CARG1, >5
+ | gettp CARG2, CARG2
+ | daddiu CARG2, CARG2, -LJ_TTAB
+ | lbu TMP1, CFUNC:CARG1->ffid
+ | daddiu CARG3, CARG3, -LJ_TNIL
+ | or AT, CARG2, CARG3
+ | daddiu TMP1, TMP1, -FF_next_N
+ | or AT, AT, TMP1
+ | bnez AT, >5
+ |. lui TMP1, 0xfffe
+ | daddu PC, TMP0, TMP2
+ | ori TMP1, TMP1, 0x7fff
+ | dsll TMP1, TMP1, 32
+ | sd TMP1, -8(RA)
+ |1:
+ | ins_next
+ |5: // Despecialize bytecode if any of the checks fail.
+ | li TMP3, BC_JMP
+ | li TMP1, BC_ITERC
+ | sb TMP3, -4+OFS_OP(PC)
+ | daddu PC, TMP0, TMP2
+ | b <1
+ |. sb TMP1, OFS_OP(PC)
+ break;
+
+ case BC_VARG:
+ | // RA = base*8, RB = (nresults+1)*8, RC = numparams*8
+ | ld TMP0, FRAME_PC(BASE)
+ | decode_RDtoRC8 RC, RD
+ | decode_RB8a RB, INS
+ | daddu RC, BASE, RC
+ | decode_RB8b RB
+ | daddu RA, BASE, RA
+ | daddiu RC, RC, FRAME_VARG
+ | daddu TMP2, RA, RB
+ | daddiu TMP3, BASE, -16 // TMP3 = vtop
+ | dsubu RC, RC, TMP0 // RC = vbase
+ | // Note: RC may now be even _above_ BASE if nargs was < numparams.
+ | beqz RB, >5 // Copy all varargs?
+ |. dsubu TMP1, TMP3, RC
+ | daddiu TMP2, TMP2, -16
+ |1: // Copy vararg slots to destination slots.
+ | ld CARG1, 0(RC)
+ | sltu AT, RC, TMP3
+ | daddiu RC, RC, 8
+ | movz CARG1, TISNIL, AT
+ | sd CARG1, 0(RA)
+ | sltu AT, RA, TMP2
+ | bnez AT, <1
+ |. daddiu RA, RA, 8
+ |3:
+ | ins_next
+ |
+ |5: // Copy all varargs.
+ | ld TMP0, L->maxstack
+ | blez TMP1, <3 // No vararg slots?
+ |. li MULTRES, 8 // MULTRES = (0+1)*8
+ | daddu TMP2, RA, TMP1
+ | sltu AT, TMP0, TMP2
+ | bnez AT, >7
+ |. daddiu MULTRES, TMP1, 8
+ |6:
+ | ld CRET1, 0(RC)
+ | daddiu RC, RC, 8
+ | sd CRET1, 0(RA)
+ | sltu AT, RC, TMP3
+ | bnez AT, <6 // More vararg slots?
+ |. daddiu RA, RA, 8
+ | b <3
+ |. nop
+ |
+ |7: // Grow stack for varargs.
+ | load_got lj_state_growstack
+ | sd RA, L->top
+ | dsubu RA, RA, BASE
+ | sd BASE, L->base
+ | dsubu BASE, RC, BASE // Need delta, because BASE may change.
+ | sd PC, SAVE_PC
+ | srl CARG2, TMP1, 3
+ | call_intern lj_state_growstack // (lua_State *L, int n)
+ |. move CARG1, L
+ | move RC, BASE
+ | ld BASE, L->base
+ | daddu RA, BASE, RA
+ | daddu RC, BASE, RC
+ | b <6
+ |. daddiu TMP3, BASE, -16
+ break;
+
+ /* -- Returns ----------------------------------------------------------- */
+
+ case BC_RETM:
+ | // RA = results*8, RD = extra_nresults*8
+ | addu RD, RD, MULTRES // MULTRES >= 8, so RD >= 8.
+ | // Fall through. Assumes BC_RET follows.
+ break;
+
+ case BC_RET:
+ | // RA = results*8, RD = (nresults+1)*8
+ | ld PC, FRAME_PC(BASE)
+ | daddu RA, BASE, RA
+ | move MULTRES, RD
+ |1:
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->BC_RETV_Z
+ |. xori TMP1, PC, FRAME_VARG
+ |
+ |->BC_RET_Z:
+ | // BASE = base, RA = resultptr, RD = (nresults+1)*8, PC = return
+ | lw INS, -4(PC)
+ | daddiu TMP2, BASE, -16
+ | daddiu RC, RD, -8
+ | decode_RA8a TMP0, INS
+ | decode_RB8a RB, INS
+ | decode_RA8b TMP0
+ | decode_RB8b RB
+ | daddu TMP3, TMP2, RB
+ | beqz RC, >3
+ |. dsubu BASE, TMP2, TMP0
+ |2:
+ | ld CRET1, 0(RA)
+ | daddiu RA, RA, 8
+ | daddiu RC, RC, -8
+ | sd CRET1, 0(TMP2)
+ | bnez RC, <2
+ |. daddiu TMP2, TMP2, 8
+ |3:
+ | daddiu TMP3, TMP3, -8
+ |5:
+ | sltu AT, TMP2, TMP3
+ | bnez AT, >6
+ |. ld LFUNC:TMP1, FRAME_FUNC(BASE)
+ | ins_next1
+ | cleartp LFUNC:TMP1
+ | ld TMP1, LFUNC:TMP1->pc
+ | ld KBASE, PC2PROTO(k)(TMP1)
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | sd TISNIL, 0(TMP2)
+ | b <5
+ |. daddiu TMP2, TMP2, 8
+ |
+ |->BC_RETV_Z: // Non-standard return case.
+ | andi TMP2, TMP1, FRAME_TYPEP
+ | bnez TMP2, ->vm_return
+ |. nop
+ | // Return from vararg function: relocate BASE down.
+ | dsubu BASE, BASE, TMP1
+ | b <1
+ |. ld PC, FRAME_PC(BASE)
+ break;
+
+ case BC_RET0: case BC_RET1:
+ | // RA = results*8, RD = (nresults+1)*8
+ | ld PC, FRAME_PC(BASE)
+ | daddu RA, BASE, RA
+ | move MULTRES, RD
+ | andi TMP0, PC, FRAME_TYPE
+ | bnez TMP0, ->BC_RETV_Z
+ |. xori TMP1, PC, FRAME_VARG
+ | lw INS, -4(PC)
+ | daddiu TMP2, BASE, -16
+ if (op == BC_RET1) {
+ | ld CRET1, 0(RA)
+ }
+ | decode_RB8a RB, INS
+ | decode_RA8a RA, INS
+ | decode_RB8b RB
+ | decode_RA8b RA
+ | dsubu BASE, TMP2, RA
+ if (op == BC_RET1) {
+ | sd CRET1, 0(TMP2)
+ }
+ |5:
+ | sltu AT, RD, RB
+ | bnez AT, >6
+ |. ld TMP1, FRAME_FUNC(BASE)
+ | ins_next1
+ | cleartp LFUNC:TMP1
+ | ld TMP1, LFUNC:TMP1->pc
+ | ld KBASE, PC2PROTO(k)(TMP1)
+ | ins_next2
+ |
+ |6: // Fill up results with nil.
+ | daddiu TMP2, TMP2, 8
+ | daddiu RD, RD, 8
+ | b <5
+ if (op == BC_RET1) {
+ |. sd TISNIL, 0(TMP2)
+ } else {
+ |. sd TISNIL, -8(TMP2)
+ }
+ break;
+
+ /* -- Loops and branches ------------------------------------------------ */
+
+ case BC_FORL:
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_IFORL follows.
+ break;
+
+ case BC_JFORI:
+ case BC_JFORL:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_FORI:
+ case BC_IFORL:
+ | // RA = base*8, RD = target (after end of loop or start of loop)
+ vk = (op == BC_IFORL || op == BC_JFORL);
+ | daddu RA, BASE, RA
+ | ld CARG1, FORL_IDX*8(RA) // IDX CARG1 - CARG3 type
+ | gettp CARG3, CARG1
+ if (op != BC_JFORL) {
+ | srl RD, RD, 1
+ | lui TMP2, (-(BCBIAS_J*4 >> 16) & 65535)
+ | daddu TMP2, RD, TMP2
+ }
+ if (!vk) {
+ | ld CARG2, FORL_STOP*8(RA) // STOP CARG2 - CARG4 type
+ | ld CRET1, FORL_STEP*8(RA) // STEP CRET1 - CRET2 type
+ | gettp CARG4, CARG2
+ | bne CARG3, TISNUM, >5
+ |. gettp CRET2, CRET1
+ | bne CARG4, TISNUM, ->vmeta_for
+ |. sextw CARG3, CARG1
+ | bne CRET2, TISNUM, ->vmeta_for
+ |. sextw CARG2, CARG2
+ | dext AT, CRET1, 31, 0
+ | slt CRET1, CARG2, CARG3
+ | slt TMP1, CARG3, CARG2
+ | movn CRET1, TMP1, AT
+ } else {
+ | bne CARG3, TISNUM, >5
+ |. ld CARG2, FORL_STEP*8(RA) // STEP CARG2 - CARG4 type
+ | ld CRET1, FORL_STOP*8(RA) // STOP CRET1 - CRET2 type
+ | sextw TMP3, CARG1
+ | sextw CARG2, CARG2
+ | sextw CRET1, CRET1
+ | addu CARG1, TMP3, CARG2
+ | xor TMP0, CARG1, TMP3
+ | xor TMP1, CARG1, CARG2
+ | and TMP0, TMP0, TMP1
+ | slt TMP1, CARG1, CRET1
+ | slt CRET1, CRET1, CARG1
+ | slt AT, CARG2, r0
+ | slt TMP0, TMP0, r0 // ((y^a) & (y^b)) < 0: overflow.
+ | movn CRET1, TMP1, AT
+ | or CRET1, CRET1, TMP0
+ | zextw CARG1, CARG1
+ | settp CARG1, TISNUM
+ }
+ |1:
+ if (op == BC_FORI) {
+ | movz TMP2, r0, CRET1
+ | daddu PC, PC, TMP2
+ } else if (op == BC_JFORI) {
+ | daddu PC, PC, TMP2
+ | lhu RD, -4+OFS_RD(PC)
+ } else if (op == BC_IFORL) {
+ | movn TMP2, r0, CRET1
+ | daddu PC, PC, TMP2
+ }
+ if (vk) {
+ | sd CARG1, FORL_IDX*8(RA)
+ }
+ | ins_next1
+ | sd CARG1, FORL_EXT*8(RA)
+ |2:
+ if (op == BC_JFORI) {
+ | beqz CRET1, =>BC_JLOOP
+ |. decode_RD8b RD
+ } else if (op == BC_JFORL) {
+ | beqz CRET1, =>BC_JLOOP
+ }
+ | ins_next2
+ |
+ |5: // FP loop.
+ |.if FPU
+ if (!vk) {
+ | ldc1 f0, FORL_IDX*8(RA)
+ | ldc1 f2, FORL_STOP*8(RA)
+ | sltiu TMP0, CARG3, LJ_TISNUM
+ | sltiu TMP1, CARG4, LJ_TISNUM
+ | sltiu AT, CRET2, LJ_TISNUM
+ | ld TMP3, FORL_STEP*8(RA)
+ | and TMP0, TMP0, TMP1
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_for
+ |. slt TMP3, TMP3, r0
+ | c.ole.d 0, f0, f2
+ | c.ole.d 1, f2, f0
+ | li CRET1, 1
+ | movt CRET1, r0, 0
+ | movt AT, r0, 1
+ | b <1
+ |. movn CRET1, AT, TMP3
+ } else {
+ | ldc1 f0, FORL_IDX*8(RA)
+ | ldc1 f4, FORL_STEP*8(RA)
+ | ldc1 f2, FORL_STOP*8(RA)
+ | ld TMP3, FORL_STEP*8(RA)
+ | add.d f0, f0, f4
+ | c.ole.d 0, f0, f2
+ | c.ole.d 1, f2, f0
+ | slt TMP3, TMP3, r0
+ | li CRET1, 1
+ | li AT, 1
+ | movt CRET1, r0, 0
+ | movt AT, r0, 1
+ | movn CRET1, AT, TMP3
+ if (op == BC_IFORL) {
+ | movn TMP2, r0, CRET1
+ | daddu PC, PC, TMP2
+ }
+ | sdc1 f0, FORL_IDX*8(RA)
+ | ins_next1
+ | b <2
+ |. sdc1 f0, FORL_EXT*8(RA)
+ }
+ |.else
+ if (!vk) {
+ | sltiu TMP0, CARG3, LJ_TISNUM
+ | sltiu TMP1, CARG4, LJ_TISNUM
+ | sltiu AT, CRET2, LJ_TISNUM
+ | and TMP0, TMP0, TMP1
+ | and AT, AT, TMP0
+ | beqz AT, ->vmeta_for
+ |. nop
+ | bal ->vm_sfcmpolex
+ |. lw TMP3, FORL_STEP*8+HI(RA)
+ | b <1
+ |. nop
+ } else {
+ | load_got __adddf3
+ | call_extern
+ |. sw TMP2, TMPD
+ | ld CARG2, FORL_STOP*8(RA)
+ | move CARG1, CRET1
+ if ( op == BC_JFORL ) {
+ | lhu RD, -4+OFS_RD(PC)
+ | decode_RD8b RD
+ }
+ | bal ->vm_sfcmpolex
+ |. lw TMP3, FORL_STEP*8+HI(RA)
+ | b <1
+ |. lw TMP2, TMPD
+ }
+ |.endif
+ break;
+
+ case BC_ITERL:
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_IITERL follows.
+ break;
+
+ case BC_JITERL:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_IITERL:
+ | // RA = base*8, RD = target
+ | daddu RA, BASE, RA
+ | ld TMP1, 0(RA)
+ | beq TMP1, TISNIL, >1 // Stop if iterator returned nil.
+ |. nop
+ if (op == BC_JITERL) {
+ | b =>BC_JLOOP
+ |. sd TMP1, -8(RA)
+ } else {
+ | branch_RD // Otherwise save control var + branch.
+ | sd TMP1, -8(RA)
+ }
+ |1:
+ | ins_next
+ break;
+
+ case BC_LOOP:
+ | // RA = base*8, RD = target (loop extent)
+ | // Note: RA/RD is only used by trace recorder to determine scope/extent
+ | // This opcode does NOT jump, it's only purpose is to detect a hot loop.
+ |.if JIT
+ | hotloop
+ |.endif
+ | // Fall through. Assumes BC_ILOOP follows.
+ break;
+
+ case BC_ILOOP:
+ | // RA = base*8, RD = target (loop extent)
+ | ins_next
+ break;
+
+ case BC_JLOOP:
+ |.if JIT
+ | // RA = base*8 (ignored), RD = traceno*8
+ | ld TMP1, DISPATCH_J(trace)(DISPATCH)
+ | li AT, 0
+ | daddu TMP1, TMP1, RD
+ | // Traces on MIPS don't store the trace number, so use 0.
+ | sd AT, DISPATCH_GL(vmstate)(DISPATCH)
+ | ld TRACE:TMP2, 0(TMP1)
+ | sd BASE, DISPATCH_GL(jit_base)(DISPATCH)
+ | ld TMP2, TRACE:TMP2->mcode
+ | sd L, DISPATCH_GL(tmpbuf.L)(DISPATCH)
+ | jr TMP2
+ |. daddiu JGL, DISPATCH, GG_DISP2G+32768
+ |.endif
+ break;
+
+ case BC_JMP:
+ | // RA = base*8 (only used by trace recorder), RD = target
+ | branch_RD
+ | ins_next
+ break;
+
+ /* -- Function headers -------------------------------------------------- */
+
+ case BC_FUNCF:
+ |.if JIT
+ | hotcall
+ |.endif
+ case BC_FUNCV: /* NYI: compiled vararg functions. */
+ | // Fall through. Assumes BC_IFUNCF/BC_IFUNCV follow.
+ break;
+
+ case BC_JFUNCF:
+#if !LJ_HASJIT
+ break;
+#endif
+ case BC_IFUNCF:
+ | // BASE = new base, RA = BASE+framesize*8, RB = LFUNC, RC = nargs*8
+ | ld TMP2, L->maxstack
+ | lbu TMP1, -4+PC2PROTO(numparams)(PC)
+ | ld KBASE, -4+PC2PROTO(k)(PC)
+ | sltu AT, TMP2, RA
+ | bnez AT, ->vm_growstack_l
+ |. sll TMP1, TMP1, 3
+ if (op != BC_JFUNCF) {
+ | ins_next1
+ }
+ |2:
+ | sltu AT, NARGS8:RC, TMP1 // Check for missing parameters.
+ | bnez AT, >3
+ |. daddu AT, BASE, NARGS8:RC
+ if (op == BC_JFUNCF) {
+ | decode_RD8a RD, INS
+ | b =>BC_JLOOP
+ |. decode_RD8b RD
+ } else {
+ | ins_next2
+ }
+ |
+ |3: // Clear missing parameters.
+ | sd TISNIL, 0(AT)
+ | b <2
+ |. addiu NARGS8:RC, NARGS8:RC, 8
+ break;
+
+ case BC_JFUNCV:
+#if !LJ_HASJIT
+ break;
+#endif
+ | NYI // NYI: compiled vararg functions
+ break; /* NYI: compiled vararg functions. */
+
+ case BC_IFUNCV:
+ | // BASE = new base, RA = BASE+framesize*8, RB = LFUNC, RC = nargs*8
+ | li TMP0, LJ_TFUNC
+ | daddu TMP1, BASE, RC
+ | ld TMP2, L->maxstack
+ | settp LFUNC:RB, TMP0
+ | daddu TMP0, RA, RC
+ | sd LFUNC:RB, 0(TMP1) // Store (tagged) copy of LFUNC.
+ | daddiu TMP3, RC, 16+FRAME_VARG
+ | sltu AT, TMP0, TMP2
+ | ld KBASE, -4+PC2PROTO(k)(PC)
+ | beqz AT, ->vm_growstack_l
+ |. sd TMP3, 8(TMP1) // Store delta + FRAME_VARG.
+ | lbu TMP2, -4+PC2PROTO(numparams)(PC)
+ | move RA, BASE
+ | move RC, TMP1
+ | ins_next1
+ | beqz TMP2, >3
+ |. daddiu BASE, TMP1, 16
+ |1:
+ | ld TMP0, 0(RA)
+ | sltu AT, RA, RC // Less args than parameters?
+ | move CARG1, TMP0
+ | movz TMP0, TISNIL, AT // Clear missing parameters.
+ | movn CARG1, TISNIL, AT // Clear old fixarg slot (help the GC).
+ | addiu TMP2, TMP2, -1
+ | sd TMP0, 16(TMP1)
+ | daddiu TMP1, TMP1, 8
+ | sd CARG1, 0(RA)
+ | bnez TMP2, <1
+ |. daddiu RA, RA, 8
+ |3:
+ | ins_next2
+ break;
+
+ case BC_FUNCC:
+ case BC_FUNCCW:
+ | // BASE = new base, RA = BASE+framesize*8, RB = CFUNC, RC = nargs*8
+ if (op == BC_FUNCC) {
+ | ld CFUNCADDR, CFUNC:RB->f
+ } else {
+ | ld CFUNCADDR, DISPATCH_GL(wrapf)(DISPATCH)
+ }
+ | daddu TMP1, RA, NARGS8:RC
+ | ld TMP2, L->maxstack
+ | daddu RC, BASE, NARGS8:RC
+ | sd BASE, L->base
+ | sltu AT, TMP2, TMP1
+ | sd RC, L->top
+ | li_vmstate C
+ if (op == BC_FUNCCW) {
+ | ld CARG2, CFUNC:RB->f
+ }
+ | bnez AT, ->vm_growstack_c // Need to grow stack.
+ |. move CARG1, L
+ | jalr CFUNCADDR // (lua_State *L [, lua_CFunction f])
+ |. st_vmstate
+ | // Returns nresults.
+ | ld BASE, L->base
+ | sll RD, CRET1, 3
+ | ld TMP1, L->top
+ | li_vmstate INTERP
+ | ld PC, FRAME_PC(BASE) // Fetch PC of caller.
+ | dsubu RA, TMP1, RD // RA = L->top - nresults*8
+ | sd L, DISPATCH_GL(cur_L)(DISPATCH)
+ | b ->vm_returnc
+ |. st_vmstate
+ break;
+
+ /* ---------------------------------------------------------------------- */
+
+ default:
+ fprintf(stderr, "Error: undefined opcode BC_%s\n", bc_names[op]);
+ exit(2);
+ break;
+ }
+}
+
+static int build_backend(BuildCtx *ctx)
+{
+ int op;
+
+ dasm_growpc(Dst, BC__MAX);
+
+ build_subroutines(ctx);
+
+ |.code_op
+ for (op = 0; op < BC__MAX; op++)
+ build_ins(ctx, (BCOp)op, op);
+
+ return BC__MAX;
+}
+
+/* Emit pseudo frame-info for all assembler functions. */
+static void emit_asm_debug(BuildCtx *ctx)
+{
+ int fcofs = (int)((uint8_t *)ctx->glob[GLOB_vm_ffi_call] - ctx->code);
+ int i;
+ switch (ctx->mode) {
+ case BUILD_elfasm:
+ fprintf(ctx->fp, "\t.section .debug_frame,\"\",@progbits\n");
+ fprintf(ctx->fp,
+ ".Lframe0:\n"
+ "\t.4byte .LECIE0-.LSCIE0\n"
+ ".LSCIE0:\n"
+ "\t.4byte 0xffffffff\n"
+ "\t.byte 0x1\n"
+ "\t.string \"\"\n"
+ "\t.uleb128 0x1\n"
+ "\t.sleb128 -4\n"
+ "\t.byte 31\n"
+ "\t.byte 0xc\n\t.uleb128 29\n\t.uleb128 0\n"
+ "\t.align 2\n"
+ ".LECIE0:\n\n");
+ fprintf(ctx->fp,
+ ".LSFDE0:\n"
+ "\t.4byte .LEFDE0-.LASFDE0\n"
+ ".LASFDE0:\n"
+ "\t.4byte .Lframe0\n"
+ "\t.8byte .Lbegin\n"
+ "\t.8byte %d\n"
+ "\t.byte 0xe\n\t.uleb128 %d\n"
+ "\t.byte 0x9f\n\t.sleb128 2*5\n"
+ "\t.byte 0x9e\n\t.sleb128 2*6\n",
+ fcofs, CFRAME_SIZE);
+ for (i = 23; i >= 16; i--)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+i, 2*(30-i));
+#if !LJ_SOFTFP
+ for (i = 31; i >= 24; i--)
+ fprintf(ctx->fp, "\t.byte %d\n\t.uleb128 %d\n", 0x80+32+i, 2*(46-i));
+#endif
+ fprintf(ctx->fp,
+ "\t.align 2\n"
+ ".LEFDE0:\n\n");
+#if LJ_HASFFI
+ fprintf(ctx->fp,
+ ".LSFDE1:\n"
+ "\t.4byte .LEFDE1-.LASFDE1\n"
+ ".LASFDE1:\n"
+ "\t.4byte .Lframe0\n"
+ "\t.4byte lj_vm_ffi_call\n"
+ "\t.4byte %d\n"
+ "\t.byte 0x9f\n\t.uleb128 2*1\n"
+ "\t.byte 0x90\n\t.uleb128 2*2\n"
+ "\t.byte 0xd\n\t.uleb128 0x10\n"
+ "\t.align 2\n"
+ ".LEFDE1:\n\n", (int)ctx->codesz - fcofs);
+#endif
+#if !LJ_NO_UNWIND
+ /* NYI */
+#endif
+ break;
+ default:
+ break;
+ }
+}
+
diff --git a/Build/source/libs/luajit/TLpatches/patch-04 b/Build/source/libs/luajit/TLpatches/patch-04
index 83881026a1d..c382ca11993 100644
--- a/Build/source/libs/luajit/TLpatches/patch-04
+++ b/Build/source/libs/luajit/TLpatches/patch-04
@@ -33,12 +33,16 @@ diff -bur LuaJIT-2.1.0-beta3-orig/src/lauxlib.h LuaJIT-2.1.0-beta3/src/lauxlib.h
/*
** ===============================================================
** some useful macros
-@@ -155,6 +180,8 @@
+@@ -155,6 +180,12 @@
LUALIB_API void (luaL_addvalue) (luaL_Buffer *B);
LUALIB_API void (luaL_pushresult) (luaL_Buffer *B);
++/* -- Luajittex needs this one because it's faster than make it Lua -- */
++LUA_API int (RESERVED_lua_dump) (lua_State *L, lua_Writer writer, void *data, int strip);
+/* Luajittex needs this one because it overloads loadfile */
+LUALIB_API int (RESERVED_load_aux_JIT) (lua_State *L, int status, int envarg);
++
++
/* }====================================================== */
diff --git a/Build/source/libs/luajit/TLpatches/patch-09 b/Build/source/libs/luajit/TLpatches/patch-09
index 312031f36f2..d0fcf9b2946 100644
--- a/Build/source/libs/luajit/TLpatches/patch-09
+++ b/Build/source/libs/luajit/TLpatches/patch-09
@@ -1,11 +1,20 @@
diff -bur LuaJIT-2.1.0-beta3-orig/src/lj_load.c LuaJIT-2.1.0-beta3/src/lj_load.c
--- LuaJIT-2.1.0-beta3-orig/src/lj_load.c 2017-05-01 21:05:00.000000000 +0200
+++ LuaJIT-2.1.0-beta3/src/lj_load.c 2017-06-19 18:05:35.864190001 +0200
-@@ -166,3 +166,22 @@
+@@ -166,3 +166,31 @@
return 1;
}
-+
++/* -- Luajittex needs this one because it's faster than make it Lua -- */
++LUA_API int RESERVED_lua_dump(lua_State *L, lua_Writer writer, void *data, int strip)
++{
++ cTValue *o = L->top-1;
++ api_check(L, L->top > L->base);
++ if (tvisfunc(o) && isluafunc(funcV(o)))
++ return lj_bcwrite(L, funcproto(funcV(o)), writer, data, strip);
++ else
++ return 1;
++}
+
+/* -- Luajittex needs this one because it overloads loadfile -- */
+LUALIB_API int RESERVED_load_aux_JIT(lua_State *L, int status, int envarg)